Wainer, Ralph Final Phd Thesis.Pdf

Soil legacy and fungal community

A thesis submitted in partial fulfilment of the requirements for the degree of Doctor of Philosophy in Biology

2020 Ralph Wainer The University of Canterbury General abstract

The goal of my thesis was to study the effects of soil under various levels of invasive Cytisus scoparius (Scotch broom) and then examine whether the unique soil legacy of C. scoparius was contingent on how C. scoparius shaped soil fungal communities.

I began my research by studying the effect of the soil legacy of C. scoparius in a controlled environment (via a greenhouse experiment; Chapter 2). Knowing the effect of the soil legacy of C. scoparius under regulated conditions, I then undertook a field survey (via a natural experiment; Chapter 3), in which I systematically recorded changes in fungal community composition across a natural density gradient of C. scoparius invasion. I subsequently investigated whether the environmental DNA (eDNA) metabarcoding techniques I applied throughout my natural survey could be optimised for future researchers (via a methodological experiment; Chapter 4). Lastly, I analysed how different fungal communities found near C. scoparius may underlie the results of my greenhouse experiment (via mixed-effect modelling; Chapter 5).

In Chapter 2, I found contrary to my hypothesis that the effects of soil extracted under various levels of C. scoparius invasion favoured the growth of native New Zealand plants over its own taxonomic family in a controlled greenhouse environment. Given that the predominantly positive soil legacy of C. scoparius could only be partly attributed to soil chemical traits, microbial effects likely played an underlying role in the invasion success of C. scoparius. In Chapter 3, I found that fungal diversity in soil under C. scoparius was unexpectedly higher than in grassland uninvaded by C. scoparius, and that C. scoparius invasion resulted in increased homogenisation of certain fungal groups within the overall soil fungal community. My results suggested that coalescence between previously separated fungal communities may have occurred due to C. scoparius invasion. Apart from C. scoparius having a definite effect on soil fungal communities, it is possible that the soil fungal communities themselves might contribute to the shrub’s invasiveness, which I further tested in a field-experiment (Appendix E). In Chapter 4, I present the pitfalls and benefits of eDNA pooling, identifying a fungal taxon-wide bias in the proportional abundance of fungi in pooled eDNA samples. I demonstrate how rarer fungi remain increasingly unaccounted for with increased degrees of pooling, yet also show how pooling may benefit researchers who wish to study the larger-scale effect of environmental drivers (e.g., anthropogenic effects, invasive species impacts). In Chapter 5, I show how increased arbuscular mycorrhizal richness found in more homogenised soil communities (studied in Chapter 1) were partly responsible for the generally positive soil legacy of C. scoparius, especially for exotic Fabaceae which can probably benefit more from arbuscular mycorrhizal fungi-facilitated P enrichment due to their ability to fix N.

By demonstrating how changes in fungal communities caused by an invasive N-fixing plant may impact plant growth and nutrient acquisition, the results of my thesis highlight the importance of incorporating fungal community composition in soil legacy studies. Although biodiversity losses of plants and other organisms following invasion are common, I show how soil fungal communities may be considered an exception to the rule. I highlight the importance of systematic sample processing and encourage the use of eDNA metabarcoding techniques to better understand how changes in soil fungal communities may possibly benefit native plants in ecological restoration projects or adversely underlie an exotic shrub’s invasiveness.

Deputy Vice-Chancellor’s Office Postgraduate Research Office

Co-Authorship Form

This form is to accompany the submission of any thesis that contains research reported in co-authored work that has been published, accepted for publication, or submitted for publication. A copy of this form should be included for each co-authored work that is included in the thesis. Completed forms should be included at the front (after the thesis abstract) of each copy of the thesis submitted for examination and library deposit.

Please indicate the chapter/section/pages of this thesis that are extracted from co-authored work and provide details of the publication or submission from the extract comes: Community-level direct and indirect impacts of an invasive plant favour exotic over native species – Appendix E

Please detail the nature and extent (%) of contribution by the candidate: 50% - The candidate is joint-first author on the published Journal of Ecology research article “Community‐level direct and indirect impacts of an invasive plant favour exotic over native species”, which is included in the Appendix of the thesis.

Certification by Co-authors:

If there is more than one co-author then a single co-author can sign on behalf of all

The undersigned certifies that:

▪ The above statement correctly reflects the nature and extent of the Doctoral candidate’s contribution to this co-authored work ▪ In cases where the candidate was the lead author of the co-authored work he or she wrote the text

Name: Warwick Allen Signature: Date: 29/06/2020

Acknowledgments

This research was funded by the Bio-Protection Research Centre (BPRC) and the National Science Challenge, with additional support from the Ross Beever Memorial Mycological Award. I thank the landowners and the Department of Conservation (DOC) for access to the study sites, Xiao Xiao Lin and the team at Massey Genome Service for their support with sequencing, and Ngaire Foster along with the team at Manaaki Whenua for their help with plant nutrient analysis. The research plots at Molesworth Station were initially established by Manaaki Whenua researchers, including D. Peltzer, M. St John, C. Morse and K. Orwin.

I am greatly indebted to my supervisory team, who allowed me to embark on this academic adventure in this beautiful part of the world: Ian, Eirian and Hayley - thank you for all your guidance and support during these last three years and for enabling me to learn many rare lessons. I wouldn’t wish for any other group of supervisors. I am likewise grateful that I could both know and work alongside Warwick Allen and all members of the Ecosystem Mycology Group. My thanks goes to all my collaborators and colleagues at the BPRC, to Lauren, Andi, Sam, Rowan, Jacopo, Andrei, Romy, Tom, Isabelle, Francesco, Aimee, Kuchar, Phil and Will.

I am thankful for Angela Wakelin’s, Brigitta Kurenbach’s and Steven Gieseg’s help in the wet-lab, Dave Conder’s cheerful assistance within the greenhouses and for the purchasing powers of both Brian Kwan and Angela Langrish. Alan Woods, Linda Morris, Jennifer Bufford and Andrew Holyoake all helped address unexpected pickles and my fieldwork and sample processing wouldn’t have been possible without the continued aid of many students, especially Georgia, Nils, Laura and Marcus-Rongowhitiao. I am glad to have been able to work alongside Joanna, Vanita, John, Tyler, Sarah, Jonathan and Zach and to have learnt so much from so many. A special shout-out goes to my academic ‘grandparents’, to Rus, Yves, Lilia, Nancy, Mia and Jerzy.

My final thanks goes to my parents, to Anna, Christina and Robert for their unending support.

Table of Contents

General abstract ...... 1 Acknowledgments ...... 4 Chapter 1: General introduction ...... 6 Chapter 2: The soil legacy of Cytisus scoparius ...... 12 Introduction ...... 13 Methods ...... 17 Results ...... 23 Discussion ...... 30 Chapter 3: The response of fungal communities to Cytisus scoparius invasion ...... 35 Introduction ...... 36 Methods ...... 41 Results ...... 46 Discussion ...... 58 Chapter 4: Consequences of environmental DNA pooling ...... 64 Introduction ...... 65 Methods ...... 69 Results ...... 75 Discussion ...... 87 Chapter 5: The fungal component of the soil legacy of Cytisus scoparius ...... 91 Introduction ...... 92 Methods ...... 94 Results ...... 96 Discussion ...... 101 Chapter 6: General discussion ...... 105 References ...... 111 Appendix A (Supplement Chapter 2) ...... 134 Appendix B (Supplement Chapter 3) ...... 139 Appendix C (Supplement Chapter 4) ...... 157 Appendix D (Supplement Chapter 5)...... 180 Appendix E – Allen et al. (2020) ...... 183

Chapter 1: General introduction

Background: The invasion of Cytisus scoparius in New Zealand

Invasions by shrubs from the family Fabaceae have occurred as a direct consequence of woodland clearance since the Bronze Age (Figueiral and Bettencourt 2004). This process has been accelerated by the world-wide transport, introduction and subsequent invasion by non-native plants (Jelbert et al. 2019, Hill et al. 2020). Between 1840 and 2000, over 8 million hectares of New Zealand’s native woodland have been replaced with grass- and shrubland (Steens et al. 2007). Such large-scale disturbances have often favoured the growth of exotic plants (Fogarty and Facelli 1999, Hierro et al. 2005, Christensen et al. 2019) and have led to New Zealand’s Department of Conservation adopting strategic plans to manage invasive Fabaceae (i.e., legumes) such as Cytisus scoparius (Owen 1998). Cytisus scoparius (Scotch broom) has colonized a diverse range of habitats worldwide (Potter et al. 2009) and in New Zealand (Syrett 2000), especially in the montane North Island and eastern South Island areas (Webb et al. 1988). In New Zealand, C. scoparius decreases the value of grazing pastures (Prévosto et al. 2004) and incurs costs upwards of $100 million per year in control measures and lost productivity (Jarvis et al. 2006, Saunders et al. 2017). Five biological control agents have already been introduced to New Zealand to stunt the expansion of C. scoparius (Syrett et al. 1999, Syrett et al. 2007, Paynter et al. 2012). It is probable that common factors known to promote plant invasion have facilitated the spread of C. scoparius, such as disturbance (Sokol et al. 2017), agricultural expansion (Chytrý et al. 2009, Mariotte et al. 2018), the use of superphosphate fertilizers on agricultural land (Smith 1992, Carter et al. 2019b), as well as the release of C. scoparius from natural predators and pathogens from its native range (Mitchell and Power 2003, Callaway et al. 2004, Dickie et al. 2017a).

The effect of C. scoparius invasion on vegetation and soil nutrient status is profound and long-lasting (Shaben and Myers 2010, Carter et al. 2019c) . Whereas C. scoparius has a life expectancy of 10-12 years in its native range (Waloff and Richards 1977), it has been found to live as long as 20 years in Australia (Rees and Paynter 1997) and it is possible that C. scoparius has an increased seed production (Rees and Paynter 1997, Paynter et al. 2016) and an increased pollination rate (Bode and Tong 2018) in exotic environments. Cytisus scoparius is capable of growing up to 3 metres in height within 2 years (Watt et al. 2003b) and the seeds of C. scoparius are able to remain dormant for over 3 years (Bossard 1993) allowing persistent seedbanks to form beneath C. scoparius populations (Magda et al. 2013). These seedbanks, along with the rapid growth of C. scoparius, make C. scoparius extraordinarily resistant to eradication, even after applying herbicide (Allen et al. 1995, Tran et al. 2016, Haubensak et al. 2020).

Classified as a drought tolerator (Harrington et al. 2018, Míguez-Montero et al. 2020), C. scoparius exhibits dry climate adaptations such as small deciduous leaves and a durable stem that is able to photosynthesize (Peterson and Prasad 1998), even at temperatures as low as 4℃ (Nilsen et al. 1993). The association of C. scoparius with N-fixing Bradyrhizobium (Weir et al. 2004) has enabled 30 - 50% of the shrub’s long-term nitrogen requirements to be met (Vadakattu and Paterson 2006). Due to these traits, C. scoparius is able to grow in areas with either high or low rainfall (Fogarty and Facelli 1999) and is known as a successful pioneer of open habitats that are both deficient in water and N (Williams 1981).

Cytisus scoparius generally increases soil N where it invades (Watt et al. 2003a, Haubensak and Parker 2004, Caldwell 2006, Grove et al. 2015) (however see Bellingham (1998)’s observation that C. scoparius led to soil denitrification) and although it has been suggested that the N-fixing properties of C. scoparius may benefit native plants by enriching soil N pools (Caldwell 2006), this benefit is often outweighed by C. scoparius competing for available resources (Watt et al. 2003b). Cytisus scoparius invasion has therefore generally been associated with a decline in native species (Shaben and Myers 2010). In the case of New Zealand, C. scoparius can often be found on previously forested pastoral and conservation land (Williams 1981, Bascand and Jowett 1982) alongside other invasive legumes, notably Ulex europaeus (gorse) (Reid 1973, Lee et al. 1986). Aside from protected conservation areas, C. scoparius frequently grows in New Zealand’s forestry plantations (Richardson et al. 1997, Tran et al. 2016, Carter et al. 2019a) where it is considered detrimental to the country’s pine industry (especially for Pinus radiata) as C. scoparius can stunt juvenile pine growth and in some cases cause seedling mortality (Watt et al. 2003b).

Belowground effects of invasive plants

Although 181 environmental weeds have been mapped in New Zealand (Howell and Terry 2016), among which C. scoparius has the 5th largest distribution, only a small proportion of introduced species become invasive in their new range (‘invasive’ sensu Gaston and Blackburn (2008)). Invasive plants are considered to be one of the most serious threats to local biodiversity and ecosystem functioning (Mooney and Hobbs 2000, Schultheis and MacGuigan 2018), as one of the most common consequences of a plant invasion is a reduction in species richness at the local scale (Powell et al. 2011, Essl et al. 2019). This loss in species richness is disproportionately rapid on island such as New Zealand (Spatz et al. 2017). Cytisus scoparius has long been the attention of ecological research yet most commonly with an aboveground focus (Sheppard and Hosking 2000, Syrett 2000, Sheppard et al. 2002, Bode and Tong 2018, Bode et al. 2019). Relatively less is known regarding the belowground consequences which accompany a plant invasion, with most research in this area taking a plant-soil feedback approach (Mangan et al. 2010, Crawford et al. 2019). Plants grown in soil from a site where C. scoparius has been removed might possibly benefit from N

7 enrichment (Caldwell 2006) or may adversely have their growth stunted by putatively allelopathic compounds secreted by C. scoparius (Grove et al. 2012, Pardo-Muras et al. 2018, Pardo-Muras et al. 2020).

Nitrogen fixation of Cytisus scoparius through rhizobial symbiosis

The presence and composition of soil microbial symbionts has a central influence on both soil nutrient composition and plant-plant interactions (Bever et al. 2010) and consequently plant diversity and community structure (Vogelsang et al. 2006, Manoharan et al. 2017, Semchenko et al. 2018). The invasiveness of Fabaceae such as C. scoparius can be partially linked to their ability to nodulate (Rodríguez-Echeverría et al. 2009). Compared to other N-fixing species, C. scoparius is only sparsely nodulated (Helgerson et al. 1984), yet can nonetheless exhibit high levels of N- fixation (Watt et al. 2003b, Pérez‐Fernández et al. 2017). As New Zealand has been geographically isolated for 80 million years (Radley 1989), it is postulated that coevolution of native legumes and nitrogen-fixing bacterial symbionts occurred in isolation from other major regions of legume evolution (Weir et al. 2004). There is a low diversity in legumes native to New Zealand, which are represented by only 34 species from four genera (Carmichaelia, Clianthus, Montigena and Sophora) (Heenan 1998; Heenan et al. 2001). Introduced legumes are likely to outcompete those native to New Zealand.

Despite the shrub’s prevalence, C. scoparius in New Zealand is more likely to suffer from a scarcity in compatible rhizobia partners as the rhizobial symbiosis is more plant-specific than mycorrhizal symbiosis (Rodríguez-Echeverría et al. 2009) (however see Keet et al. (2017) regarding how rhizobial symbiosis is not essential for plant invasiveness). Although a scarcity in compatible rhizobia partners has been linked to worldwide establishment or dispersal barriers across multiple legume species (Simonsen et al. 2017), legumes which can associate with a broad range of rhizobia are more likely to overcome these barriers (e.g., Klock et al. 2015). In New Zealand, all rhizobia found in introduced legumes are from the genus Bradyrhizobium, whereas most native legumes associate with the genus Mesorhizobium (Weir et al. 2004). It is likely that Bradyrhizobium was introduced to New Zealand alongside exotic legumes (Warrington et al. 2019).

Fungal diversity in New Zealand

Although the study and preservation of fungal biodiversity has garnered less attention compared to other eukaryotes, fungi encompass a substantial amount of global diversity, with fungal diversity estimates ranging from 1.5 to over 165 million species (Hawksworth 2012, Taylor et al. 2014, Tedersoo et al. 2014, Larsen et al. 2017). Censuses of soil fungi have revealed that they are highly diverse, undergo fine-scale niche partitioning (Taylor et al. 2014), and that a large proportion is saprotrophic (i.e., decomposers) (Nguyen et al. 2016). Rare fungi, particularly ectomycorrhizal

8 fungi, are sensitive to the processes of disturbance (Dickie and Reich 2005, Dickie et al. 2009) and N deposition (Avis et al. 2008), which both accompany C. scoparius invasion in New Zealand.

Fungal diversity is of integral as well as economic and cultural importance. Rare fungal species may be important sources of future pharmaceuticals (Chen et al. 2019b). In New Zealand, fungi are used by Māori as material to create traditional tattoos (Fuller et al. 2004, Dickie et al. 2020), have the potential to be used as biocontrol agents (Kuchár et al. 2019, Ehlers et al. 2020) and are deemed important to the country’s wine (Knight et al. 2020) and food industry (Guerin-Laguette et al. 2020).

Belowground effects of invasive plants: Arbuscular mycorrhizal fungi

Among the major taxa of soil fungi, the sub-phylum Glomeromycotina has a world-wide distribution (Öpik et al. 2006) and is an important component of belowground microbial communities, with ~80% of land plants being dependent on Glomeromycotina (Davison et al. 2015), including most invasive woody plants (Rejmánek and Richardson 2013). Along with some members of the sub-phylum Mucoromycotina (Orchard et al. 2017, Walker et al. 2018), Glomeromycotina are also known as arbuscular mycorrhizal fungi (AMF). Despite AMF’s global prevalence, there are only ~250 morphologically defined and 350 to 1000 molecularly defined AMF species (Davison et al. 2015). Dependent on C from plant roots, AMF confer different benefits to various plant species (Klironomos 2003, Lefebvre 2019), mainly by facilitating nutrient uptake, particularly P, although AMF are functionally quite diverse (Munkvold et al. 2004, Rivero et al. 2018) and may increase a plant’s tolerance to drought and root pathogens. Granted that most AMF are in fact mutualists (Pringle et al. 2009), there is substantial variation as to what extent AMF benefit hosts, and some plant-AMF associations have been known to reduce plant fitness (Jones and Smith 2004, Hoeksema et al. 2018, Dierks et al. 2019). It has become apparent that AMF communities are not random assemblages (Davison et al. 2011). Distinct communities of AMF are frequently formed around a given plant species (Van den Koornhuyse et al. 2003) which can impact the composition of aboveground plant communities (Tedersoo et al. 2020) in a “bottom- up” manner (Hartnett and Wilson 1999, Van der Heijden and Horton 2009).

Due to the relatively low host-specificity of AMF (Nuñez and Dickie 2014), many arbuscular mycorrhizal plants may not require AMF from their native range, but can integrate into existing ecological networks by forming novel associations with native AMF (Brundrett 2009, Zhang et al. 2017). AMF association can be an important pathway through which invasive plants alter the performance of native species (Hawkes et al. 2006, Menzel et al. 2017). It is known that a host plant changing its species of AMF has the potential to impact decomposition rates (Hodge et al. 2001, Gui et al. 2017) and plant nutrient uptake (Cavagnaro et al. 2005, Ingraffia et al. 2019) among other effects. Therefore, should an invasive plant alter an AMF community, this could potentially lead

9 to broader changes in general soil microbial community composition and structure, as well as affect resource availability and nutrient cycling (Asner et al. 2008, Rascher et al. 2011).

Belowground effects of invasive plants: Soil pathogens

The benefits of AMF can be outweighed by an opposing process such as the presence of soil pathogens. As an example, it was found that the application of fungicide to natural populations of Vulpia ciliata had no effect on plant performance (Newsham et al. 1995), which was due to a simultaneous reduction of AMF and pathogenic fungi in roots. Belowground pathogens can greatly reduce reproduction, growth and survival of plants (Burdon 1987, Raaijmakers and Mazzola 2016) and many pathogens may be host-specific (Van der Putten et al. 2007a, Bakker et al. 2018). Despite soil pathogens playing a major role in structuring plant community compositions (Latz et al. 2016), their inclusion in ecosystem studies has often been overlooked (Beckerman and Petchey 2009), which can lead to an ecosystem’s complexity being underestimated (Wood et al. 2007). Soil pathogens not only affect host plants, but also interact with other soil microorganisms (Spagnoletti et al. 2017).

One main reason why invasive plants can benefit from being introduced in a new environment is that the plants may encounter fewer host-specific pathogens compared to their native range (Reinhart and Callaway 2006, van der Putten et al. 2016). For example, it was found that the grass Ammophila arenaria, which is native to Europe, encountered fewer nematode species in New Zealand relative to its native range (Van der Putten et al. 2005). This may not always be the case, as the presence of strong indigenous belowground pathogens may also form a barrier to invasion (Reinhart and Callaway 2006). Such strong indigenous pathogens could account for a large number of seldom studied failed plant invasions and unsuccessful attempts to grow agricultural crops (Zenni and Nuñez 2013). There is limited information on soil pathogens closely associated with C. scoparius, although Cytisus sp. in North America have been observed with known soil pathogens such as Pythium sp. and Rhizoctonia sp. (Farr et al. 1989).

Use of eDNA metabarcoding in community analysis

Detecting often cryptic belowground mutualists has traditionally been difficult (Atkins and Clark 2004) reliant on specialist taxonomic knowledge of belowground biota (McCartney et al. 2003). In contrast, using molecular-based detection methods to identify soil biota yields more reproducible and rapid results (Aslam et al. 2017, Nilsson et al. 2019a) (although see Malarczyk et al. (2019) regarding alternative methods). An increasingly popular molecular method is metabarcoding, which combines DNA-based identification with high-throughput sequencing, allowing numerous species within an environmental sample to be assessed (Taberlet et al. 2012). To address the increase in metabarcoding-based studies of environmental DNA (eDNA), several methodological

10 reviews have been published on the molecular and bioinformatical steps involved (e.g., Hiraoka et al. (2016), Lear et al. (2018)), yet comparatively less attention has been given to the systematic and reproducible processing of eDNA samples prior to sequencing (Dickie et al. 2018).

Thesis aim and research objectives

The goal of my thesis was to study the effects of soil under various levels of C. scoparius invasion and then examine whether the unique soil legacy of C. scoparius was contingent on how C. scoparius shaped fungal communities in soil. Knowing the putative cause of the soil legacy of C. scoparius, I then analysed whether a plant’s direct and indirect responses to C. scoparius are observable outside of the greenhouse environment under field conditions.

I began my research by studying the effect of the soil legacy of C. scoparius in a controlled environment (via a greenhouse experiment). Knowing the effect of C. scoparius’ soil legacy under regulated conditions, I then undertook a field survey, in which I systematically recorded changes in fungal community composition across a natural density gradient of C. scoparius invasion. I subsequently studied how different fungal communities found near C. scoparius may underlie the results of my greenhouse experiment. Knowing how changes in the composition of soil fungi play a significant role in the soil legacy of C. scoparius, I measured the importance of the belowground associations of C. scoparius by examining the growth of plants in the presence of live C. scoparius (via a field-based experiment). As a final step, I investigated whether the eDNA metabarcoding techniques I used could be optimised for future researchers (via a methodological experiment).

The objectives of my thesis are encompassed in five sections:

• Via a greenhouse experiment, to study the soil legacy of C. scoparius on plants with varying species traits to determine whether C. scoparius invasion facilitates the spread of other introduced plants and/or species from the taxonomic family of C. scoparius (Chapter 2). • Conduct a field survey to obtain insight into specific changes to fungal communities induced by C. scoparius belowground, which may contribute to the invasion success of C. scoparius (Chapter 3). • Perform a methodological experiment examining the benefits and downsides of pooling metabarcoded fungal eDNA samples (Chapter 4). • Based on the results of Chapters 2 & 3, examine via mixed-effect modelling whether the influence of the soil legacy of C. scoparius on other plants may be attributed to specific fungal communities (Chapter 5). • Undertake a field-based experiment to quantify the relative importance of the direct and indirect components of the belowground impact of C. scoparius on plant growth (Allen et al. (2020); Appendix E).

Chapter 2: The soil legacy of Cytisus scoparius

Abstract

Plant invasion can cause changes in soil community composition and soil nutrient availability which may drive the composition of subsequent plant communities, particularly when the invasive plant is able to fix nitrogen. The effect of plant invasion can persist after removing the introduced plant and may be specific to individual species, which in turn may affect the success of ecological restoration projects. Using soil extracted from across a density gradient of exotic Cytisus scoparius (Fabaceae), I tested whether the shrub’s soil legacy favoured the growth (dry biomass) and nutrient acquisition (shoot % N and shoot % P) of a selection of plants native and exotic to New Zealand, which were either able or unable to fix nitrogen. I found that, compared with uninvaded soil, plants grown in C. scoparius-invaded soil had 1) higher above- and belowground biomass and higher total N:P ratios, particularly for native plants, 2) lower root:shoot ratios, 3) no apparent changes in shoot % P, and 4) a mixed response concerning shoot % N, where C. scoparius coverage did not change shoot % N in any tested Fabaceae species, yet affected half of the tested non-Fabaceae. The soil legacy of C. scoparius in a controlled greenhouse environment favoured the growth of non- leguminous native New Zealand plants over its own taxonomic family, despite prior field studies suggesting the opposite. Having found that the predominantly positive soil legacy effect of C. scoparius could only be partly attributed to soil chemical traits, microbial effects likely play an important role in the invasion success of C. scoparius.

Keywords

Cytisus scoparius, facilitation, functional traits, invasive species, removal effects, seedling establishment

Introduction

During early development, plant seedlings interact primarily with a soil environment moulded by past generations of plants. These older established plants would have interacted with and shaped communities of soil microorganisms which in turn interact with the newly recruited seedlings (Bever et al. 2010, Kulmatiski and Beard 2011). Soil microorganisms, such as fungi or bacteria, often have significant positive and negative effects on plants through root-rhizosphere mutualism (Brundrett 1991), pathogen effects (Packer and Clay 2000, Latz et al. 2016) and by driving nutrient cycles (Horwath 2017).

Plant-soil feedback, the process whereby a plant’s effect on soil community and the environment influences the growth of future generations of plants, is generally considered to be an important contributing cause of plant rarity and invasiveness in communities (Klironomos 2002). For a given plant species, soil obtained from closely related plants generally has a more negative effect on plant growth than soil obtained from distantly related plants (Kempel et al. 2018). A plant’s invasive status often correlates with more positive (or less negative) levels of plant-soil feedback (Kulmatiski et al. 2008), although both positive and negative plant-soil feedbacks have been reported for invasive plants (Reinhart et al. 2003, Suding et al. 2013, Dostálek et al. 2016, Aldorfová et al. 2020).

Positive plant-soil feedbacks can occur when certain plants accumulate specific microorganisms near their roots that benefit the plant, such as N-fixing bacteria or arbuscular mycorrhizal fungi. It has been observed that positive feedbacks lead to a loss of local community diversity (Bever et al. 1997, Bever 2002), although it has likewise been proposed that positive feedbacks are unlikely to cause a loss of diversity (Dickie et al. 2014). One form of negative plant-soil feedback occurs when pathogens accumulate in the rhizosphere of plant species. Pathogen accumulation in non-native plant species usually increases with residence time (Diez et al. 2010, van Kleunen et al. 2018). Negative plant-soil feedbacks can enhance plant community diversity by increasing species turnover rates (Klironomos 2002, Teste et al. 2017). According to a meta-analytical review by Kulmatiski et al. (2008), plant-soil feedbacks most commonly have medium to large negative effects on plant growth (with the scale of negative effects outweighing that of positive effects).

Cytisus scoparius (Scotch broom) is a leguminous shrub difficult to control with herbicides (Tran et al. 2016, Haubensak et al. 2020) which has colonized a diverse range of habitats worldwide (Holm et al. 1997, Brandes et al. 2019) and in New Zealand (Parsons and Cuthbertson 1992, Tran et al. 2016). In New Zealand, C. scoparius can often be found in low-intensity grazing grasslands alongside other invasive legumes, notably gorse (Ulex europaeus) (Lee et al. 1986, Ghanizadeh and Harrington 2019). Large scale invasions by species from the family Fabaceae have occurred as a direct consequence of woodland clearance since the dawn of agriculture (Figueiral and Bettencourt

2004). Such an invasion is happening in New Zealand at substantial economic costs (Saunders et al. 2017) and it is probable that common factors known to promote plant invasion have facilitated the spread of C. scoparius, such as disturbance (Sokol et al. 2017, Daniels and Larson 2020), agricultural expansion (Mariotte et al. 2018), the use of superphosphate fertilizers on agricultural land (Carter et al. 2019b) as well as the release of C. scoparius from fungal and viral pathogens from its native range (Mitchell and Power 2003, Dickie et al. 2017a). Cytisus scoparius invasion is associated with an increase in exotic species and/or a decline in native species, which correlates with the ability of C. scoparius to modify soil nutrient availability (Shaben and Myers 2010, Carter et al. 2019c). Given the long-lasting effects that C. scoparius invasion may have on surrounding soil, the shrub’s belowground impact on New Zealand plants deserves more attention.

Nitrogen (N) and phosphorus (P) are the two most limiting nutrients in terrestrial ecosystems (Han et al. 2005, Vitousek et al. 2010). Although minor soil denitrification caused by C. scoparius has been documented (Carter et al. 2019c), the species is commonly known as an N-fixer and can fix as much as 111 kg N ha−1 per year into aboveground tissues (Watt et al. 2003b). Once C. scoparius is removed from a site, soil affected by C. scoparius might act as a nursery plant for other species, specifically those which are unable to fix their own N. In addition to fixing N, C. scoparius increases soil organic C (Fogarty and Facelli 1999) and soils under C. scoparius have been found to have higher activities of two soil enzymes which are involved in P-acquisition (Caldwell 2006). Increases in the availability of soil P have been associated with C. scoparius invasion (Dewar et al. 2006), yet decreases in available P have likewise been recorded (Shaben and Myers 2010, Slesak et al. 2016).

The effect of the soil moulded by C. scoparius on other plants has been shown to be species-specific. On one hand, the N-fixing properties of C. scoparius might have enabled the spread of other invasive species such as hawthorn (Crataegus monogyna) (Williams et al. 2010) and C. scoparius presence has been associated with an increase in the cover of exotic sweet vernal grass (Anthoxanthum odoratum) (Carter et al. 2019c). On the other hand, C. scoparius, whether live or removed, is known to decrease the abundance of ectomycorrhizal fungi leading to an overall negative effect on seedling growth of certain pine (Pseudotsuga menziesii) (Grove et al. 2012). Assuming longer-term persistence of the effect of C. scoparius within this chapter, I refer to soil conditioned by C. scoparius as being soil representative of the soil legacy of C. scoparius.

The design of soil legacy studies requires careful handling. Studies with natural gradient approaches (i.e., which test various levels of plant invasion) are needed to better forecast how short-term plant-soil feedbacks can expand from species-level to long-term ecosystem dynamics (Kardol et al. 2012, Kardol et al. 2013). Although plant-soil feedbacks have been assumed to affect plant growth in a linear manner (Bever et al. 1997, Bever 2003), non-linear plant-soil feedback

14 effects can be overlooked if linearity is assumed when designing an experiment (Hawkes et al. 2013). All published studies on the effects of C. scoparius’ soil legacy revolve around a small number of plants and soils and none look at how different degrees of C. scoparius invasion affect plant growth.

Among the studies on the soil legacy of C. scoparius, Haubensak and Parker (2004) found that C. scoparius “may have inhibitory effects on some plants growing in invaded soils”, yet their study was restricted to a single plant (Achillea millefolium). A further study by Grove et al. (2015), which reported negative C. scoparius soil legacy, was also restricted to Pseudotsuga menziesii. Other studies have examined C. scoparius’ soil legacy, yet not in context with plant growth. Davis (2018) studied C. scoparius soil legacy using two different soils and only one plant (C. scoparius) and found that C. scoparius seedlings grew smaller in soil with C. scoparius legacy.

In a field-based experiment (Allen et al. (2020); Appendix E), we found that live C. scoparius facilitated the growth of both native and exotic legumes, while slightly favouring exotic legumes. Here I aimed to examine how both native and exotic plants respond to C. scoparius’ soil legacy and whether changes in soil composition induced by an invasive N-fixing plant might potentially promote the growth of co-invading plants from its own taxonomic family (Fabaceae) post removal. I also aimed to examine whether changes in soil composition induced by an invasive N-fixing plant might potentially promote the growth of non-leguminous plants, a functional group which was not included in our field experiment (Allen et al. (2020); Appendix E). Here I specifically use an approach which observes the growth of plants grown in soil with differing degrees of C. scoparius soil legacy, rather than a more common presence/absence approach where a plant is grown in either a ‘Control’ or ‘Effect’ soil.

I hypothesized that:

• Cytisus scoparius has soil legacies influencing the growth of other plants which differ according to plant species.

• The soil legacy of C. scoparius will have a more positive effect on exotic plants than on natives.

• As C. scoparius frequently co-occurs with other invasive members of the Fabaceae (notably Ulex europaeus), that C. scoparius’ soil legacy will have a more positive effect on members of its own taxonomic family when compared to unrelated plants.

To test these hypotheses, I use dry biomass (root and shoot) and nutrient analysis data (shoot % N and % P) from plants of varying functional groups (half native to New Zealand and half from the family Fabaceae) which were grown in soils collected from across a C. scoparius density gradient.

With accompanying data on soil chemical composition, I will have a better ability to tease apart the chemical and biological properties underlying C. scoparius’ soil legacy.

Methods

Study site and soil collection

I carried out the experiment in a single glasshouse at the University of Canterbury (Christchurch, New Zealand). I collected soil from 18 permanent sampling plots in the Saint James Conservation Area at Molesworth Station, located in the Hurunui district of the South Island of New Zealand. Each permanent sampling plot was modelled on field protocols outlined by Hurst and Allen (1993) and was previously laid out by Manaaki Whenua. The same plots have been used in previous experiments (Broadbent et al. 2017) and data on soil chemistry is available for all plots. Cytisus scoparius is widely spread throughout this region and the chosen plots range from areas uninvaded by C. scoparius to near mono-dominant C. scoparius patches. All plots were situated within 2.5 km of each other at an altitude 872-933 m above sea level. A map pinpointing the plots is shown in Figure 1 (plot coordinates are listed in Appendix A1). I collected data on C. scoparius density for each plot, including C. scoparius coverage estimates and the distance from the extracted soil to the closest mature C. scoparius.

During a 10 day period in November 2017, I extracted 25 L of the uppermost 150-200 mm of soil close to the centre of each of the 18 plots using spades sterilized in a 10% v/v bleach solution (8% sodium hypochlorite in undiluted bleach) for >10 minutes (Prince and Andrus 1992). The exact location of the collected soil from each plot is shown in Figure 2. The soils were kept in separate clean plastic bags according to plot. Subsequent soil processing at the University of Canterbury was performed within 2 weeks to minimize changes in soil microbial composition which occurs during prolonged soil storage, thus encouraging a ‘living’ effect of the soil on plants. Each soil was broken up with bleach-sterilized spading forks and all stones above 20 mm in diameter were manually removed as well as any large woody roots and plants. I then mixed the loosened soil at a 1:1 volume ratio with washed river sand. This mixing was done to minimize the quantity of soil required (and thereby any disruption caused to the permanent sampling plots) as well as to help standardize soil porosity. The soil-sand mixtures were used to fill bleach-sterilized ~1.4 litre pots with drainage holes (110 mm length × 110 mm width × 120 mm height), each containing 200-250 g of stone chips (10 mm - 20 mm) at the bottom of the pots to improve drainage.

Plant selection and glasshouse experimental layout

Sixteen plant species were used in this experiment (Table 1), each planted into each of the 18 collected soils giving a total of 288 samples. Half the plant species were native to New Zealand, the other half introduced to New Zealand. Half of the plant species are members of the Fabaceae, whereas the other half were chosen from four taxonomic families (Myrtaceae, Pinaceae, Poaceae

17 and Podocarpaceae). The selected plants can associate with arbuscular mycorrhizal fungi and ectomycorrhizal fungi, and in some cases with both (e.g., Leptospermum scoparium) (McKenzie et al. 2006) or neither (Lupinus arboreus) (Oba et al. 2001). Prior to being planted in their respective soils, some of the plants underwent stratification and/or scarification (according to the seed supplier’s recommendations) to improve germination rates. Except for the exotic grasses Agrostis capillaris and Anthoxanthum odoratum, all plants were germinated in propagation trays filled with a 1:1 mixture of vermiculite and perlite under greenhouse conditions.

During transfer of seedlings into the collected soils, an additional step was taken to account for differences in unique biota which could potentially be present in different propagation trays. Two live plant specimens and ~20g vermiculite-perlite mixture were taken from each propagation tray and mixed with tap water. The resulting slurry was then added in equal amounts to each pot, so that each sample was inoculated with the same propagation tray biota at the beginning of the growth period.

Due to the fast-growing nature of A. capillaris and A. odoratum, seeds of these grasses were directly sown on the prepared soils. For A. capillaris, it was difficult to reliably separate between individual plants, hence all plants that were not within a 30 mm diameter circle in the pot’s centre were weeded out. For A. odoratum, most pots yielded three to five distinct grasses, all of which were weeded out of the pot within the first 2 weeks except for one randomly chosen individual. Two Trifolium repens seedlings were initially planted in each pot. This additional step was done due to the delicate consistency of the seedlings, which could be damaged when transplanted from the vermiculite-perlite propagation trays into the soil pots. In the case where both T. repens seedlings survived the first two weeks in soil, one randomly chosen shoot was pulled out with its roots.

I laid out all pots on a single bench within a single glasshouse in a completely random design with ~100 mm between pots (Figure 3). Each pot stood on a ~20 mm tall upturned pot saucer which both aided water drainage and insured that water passing through the pots could not contaminate adjacent pots. The pots remained in the same position throughout the course of the experiment and were watered regularly using an overhead mist-propagation system. Anti-drip nozzles were used to prevent falling water drops splashing wet soil from one pot into another, potentially causing an unwanted cross-contamination of soil biota or minerals. Although early plant mortality proved low, any obviously dead plants were replaced within the first 6 weeks of the experiment. Plants that died after six weeks were not replaced. All pots were weeded every week throughout the growing period, which lasted ~7 months from early December 2017 to harvest in the beginning of June 2018.

The height of the plants was measured in early December immediately after planting, then in early March and lastly in late May prior to harvest. The December height measurement for A. capillaris

18 and A. odoratum was taken as zero, as seeds for these plants were directly sown onto soil and had just started germinating. Throughout the course of the experiment, no obvious plant herbivory was noticed during regular weeding, although a few plants showed minor chlorosis.

Plant harvesting and nutrient analysis

Live plants (n = 277) were destructively harvested in order to obtain above- and belowground dry biomass in the beginning of June 2018. Any deceased plants (n = 11) were excluded from further analysis. Harvesting was undertaken by washing plant roots in a basin of running water and drying the above- and belowground plant biomass in separate paper bags at 55°C for a minimum of 72 hours in aerated laboratory drying ovens prior to weighing. For A. capillaris, which was difficult to grow individually, the above- and belowground biomass of the tallest individual was measured. The exposure time of dried plants was kept to a maximum of 15 minutes when taking them out of the drying ovens for weighing, thus minimizing changes in weight caused by hydration.

Roots obtained from certain plant species (A. odoratum and A. capillaris in particular) and certain soils (e.g., plot MW17) proved difficult to wash appropriately and still contained some soil material entangled in their roots when their dry biomass was measured. For each plant, the presence of other material entwined in the roots was noted as ‘None’ (91.7%), ‘Minor’ (3.6%) or ‘Major’ (4.7%).

Of the 277 harvested plants, all shoots weighing >0.05 g (n = 262) were sent for P and N quantification via the Kjeldahl method (Bradstreet 1954) at Manaaki Whenua’s Environmental Chemistry Laboratory in Palmerston North, New Zealand. Shoots weighing <0.05 g (9% of total samples, n = 26) did not have the minimum amount of biomass required for accurate nutrient analysis and were therefore discarded. Shoots weighing >0.5g (40.6% of total samples, n = 117) were mechanically ground into a fine powder at the University of Canterbury. I pulverised all available aboveground biomass (i.e., the whole shoot) using a Retsch® MM301 mixer mill fitted with cleaned 50 mL grinding jars for 30 seconds at 30Hz. Shoots which were too large to be crushed in a single jar were ground throughout the course of several sessions, then combined. Shoots weighing 0.05-0.5 g were small enough to be ground by hand at Manaaki Whenua prior to nutrient analysis.

Statistical analysis

I used R version 3.5.0 (Team 2013) for creating graphs and conducting analyses alongside the R package “dyplr” (Wickham et al. 2015). I excluded plants found dead at harvest (3.8% of total) and when calculating root and total biomass, I only used samples with completely clean roots (91.7% of live plants). I chose to use clean roots only, as the correlation between root mass and shoot mass of all harvested live plants (R2 = 0.0108; P < 0.0001) increased markedly after removing dirty roots

(R2 = 0.2747; P < 0.0001). To quantify the effects of soil under various levels of C. scoparius invasion on plant biomass and nutrient composition according to plant species traits, I used linear mixed- effect models via the R package “lme4 (v1.121)” (Bates et al. 2014), setting plant origin, legume status and ectomycorrhizal status as fixed effects and sampling plot and plant species as random effects. The mixed models explicitly separate variation between-plant species and within-plant species, thus enabling C. scoparius’ effect on naturally larger plants to be compared with naturally smaller plants (Millar and Anderson 2004). I log-transformed the measurements I collected for “distance from extracted soil to closest C. scoparius”.

Figure 1. ArcGIS image of 18 plots in Molesworth used for soil collection (© ArcGIS, Environmental Systems Research Institute). The yellow areas in the image are predominantly populated by C. scoparius.

Figure 2. Position of soil sample extracted from each sampling plot (red triangle). Cytisus scoparius coverage estimates were measured throughout the 20 × 20 m plot as well as the distance from the extracted soil sample to the closest C. scoparius.

Table 1. List of plant species used and their status as native or introduced to New Zealand. AMF = Arbuscular mycorrhizal fungi. ECM = Ectomycorrhizal fungi. Seeds were collected in person or bought from NZ Seeds (www.nzseeds.co.nz) or Topseeds (www.topseedsltd.com). Lupinus arboreus has been considered as non-AMF (Oba et al. 2001). Leptospermum scoparium has been known to form both AMF and ECM associations (McKenzie et al. 2006). Although Pinus radiata and Pseudotsuga menziesii are primarily ectomycorrhizal, both can form AMF associations (Teste et al. 2020).

Scientific Name Common Name Fabaceae Plant Class AMF

Introduced Cytisus scoparius Scotch broom ✓ Legume ✓ to N.Z. Trifolium repens White clover ✓ Legume ✓ Ulex europaeus Gorse ✓ Legume ✓ Lupinus arboreus Yellow bush lupine ✓ Legume  Agrostis capillaris Browntop  Grass ✓ Anthoxanthum odoratum Sweet vernal grass  Grass ✓ Pinus radiata Monterey pine  Tree Trace (+ECM) Pseudotsuga menziesii Douglas Fir  Tree Trace (+ECM)

Native Sophora microphylla Kōwhai ✓ Legume ✓ to N.Z. Clianthus puniceus Kaka beak ✓ Legume ✓ Sophora tetraptera Large-leaved kōwhai ✓ Legume ✓ Carmichaelia australis N.Z. broom ✓ Legume ✓ Chionochloa conspicua Tussock grass  Grass ✓ Poa colensoi Tussock  Grass ✓ Podocarpus totara Tōtara  Tree ✓ Leptospermum scoparium Mānuka  Shrub ✓ (+ECM)

Figure 3. Set-up of the experiment at a University of Canterbury glasshouse. Each individual pot was elevated above the table to avoid cross-contamination of soil biota after mist-watering.

Results

In total, 277 live plants were harvested. When grouped according to plant origin and species functional group, exotic Fabaceae had both the largest shoot mass and tallest height (Table 2).

Effect of C. scoparius coverage on biomass

There was a significant three-way interaction for shoot biomass between C. scoparius coverage, plant origin, and legume status as well as C. scoparius coverage, plant origin and ectomycorrhizal status (Table 3; accompanying t-values in Appendix A2). With the exception of Sophora microphylla, the other seven plant species native to New Zealand showed greater aboveground dry biomass when grown in soil with C. scoparius legacy (Figure 4). The soil legacy of C. scoparius only favored half of the introduced plant species. All non-N-fixing native New Zealand plants in the experiment (n = 4) showed increased biomass under the influence of C. scoparius’ soil legacy, compared with only 1 out of 4 exotic non-N-fixing species. In general, plants from the family Fabaceae only slightly favoured growing in soil with C. scoparius legacy, with 6 out of 8 species showing increased aboveground dry biomass compared with 5 out of 8 plant species which do not fix N.

When substituting C. scoparius coverage with the log-transformed distance to the closest mature C. scoparius, there were some minor qualitative differences in results (Appendix A3), yet all 3-way interactions remained significant. Similar species trait-specific increases in biomass over C. scoparius coverage could also be observed for dry root biomass, total dry biomass and plant height at harvest (Appendix A4) and there was little qualitative difference in results when C. scoparius % coverage was substituted with distance from soil extracted to closest mature C. scoparius (Appendix A5). The inclusion of either deceased or replaced plants had a negligeable effect on results. The effect of C. scoparius’ soil legacy on shoot biomass was generally most pronounced when soil was extracted within ~2 metres of mature C. scoparius.

There was a significant one-way interaction for log-transformed root:shoot (g) ratio and C. scoparius coverage (t = -3.094; P = 0.0019). As C. scoparius coverage increased, root:shoot (g) ratio decreased for 2 native plants (i.e., Podocarpus totara, R2 = 0.2117; P = 0.0481; and Sophora tetraptera, R2 = 0.2765; P = 0.0146).

Effect of C. scoparius coverage on shoot % N and shoot % P

No significant effect was observed when analysing shoot % P over C. scoparius coverage. There was a significant three-way interaction for shoot % N between C. scoparius coverage, plant origin, and legume status (t = 2.464; P = 0.0137; Table 3). Results for shoot % N over C. scoparius coverage

23 varied. There were no significant effects for any Fabaceae, yet half of the tested non-Fabaceae reacted both positively and negatively to C. scoparius. Cytisus scoparius coverage increased shoot % N of Agrostis capillaris, Pinus radiata and Podocarpus totara, yet decreased shoot % N of Anthoxanthum odoratum (Figure 5).

There was a significant two-way interaction for shoot N:P (%) ratio between C. scoparius coverage and plant origin (t = -3.320; P = 0.0009; Table 3). Shoot N:P (%) ratio over C. scoparius coverage decreased for 2 out of 8 native plants and increased for a single exotic plant (Figure 6). Shoot N:P (%) ratios for exotic Fabaceae were notably above other plants.

Effect of C. scoparius coverage on total shoot N:P

There was a significant three-way interaction for total shoot N:P ratio between C. scoparius coverage, plant origin and legume status (t = -4.907; P < 0.0001; Table 3) as well as between C. scoparius coverage, plant origin and ectomycorrhizal status (t = 2.848; P = 0.0044; Table 3). With the exception of Clianthus puniceus, all plants which had shown a significant increase in aboveground biomass as C. scoparius coverage increased (i.e., Figure 4) likewise showed a significant increase in total shoot N:P ratio over C. scoparius coverage (Figure 7).

Effect of soil chemistry

Cytisus scoparius coverage had no detectable effect on soil chemistry (Appendix A6). In some instances, there was an effect of soil Ca (cmol(+)/kg) and soil Mg (cmol(+)/kg) on plant morphology, yet for all tested plant response variables, the effect of C. scoparius coverage superceded that of soil chemistry, both in terms of t-value and significance (Table 4). No effect was found for shoot % N over soil N(%). Uniquely for Ulex europaeus, there was a significant positive effect (post Bonferroni correction) of soil Olsen P (mg/kg) on shoot % P (R2 = 0.6801; P < 0.0001). Also uniquely for Ulex europaeus, there was a significant negative effect (post Bonferroni correction) of soil Olsen P (mg/kg) on shoot N:P (%) ratio (R2 = 0.6011; P < 0.0001) and a significant negative effect of soil K (cmol(+)/kg) on shoot N:P (%) ratio (R2 = 0.4792; P = 0.0012).

Table 2. Overview of mean plant measurements (± standard errors) for each individual species and all plants (n = 277) grouped according to species functional groups (see Table 1). AMF = Arbuscular mycorrhizal. ECM = Ectomycorrhizal. Across all examined plant species, C. scoparius on average had the highest shoot height (535.6 ± 60.27 mm), largest shoot mass (2.240 ± 0.42 g), highest total shoot N (5.22 ± 0.86 gN) and highest total shoot P (0.26 ± 0.04 gP). Cytisus scoparius’ total shoot N and total shoot P were markedly above the average of all other exotic Fabaceae (total shoot N = 2.83 ± 0.44 gN; total shoot P = 0.17 ± 0.03 gP), which in turn were above all native Fabaceae (total shoot N = 1.15 ± 0.2 gN; total shoot P = 0.10 ± 0.01 gP). Plants which solely formed arbuscular mycorrhyzal associations had a 2.79 times higher mean total N (1.59 ± 0.16 gN) compared with ectomycorrhyzal plants (Leptospermum scoparium, Pinus radiata, Pseudotsuga menziesii; mean total N = 0.57 ± 0.09 gN).

Number of live Shoot Shoot plants harvested height (mm) mass (g) %N %P Agrostis capillaris 18 174.8 ± 11.09 0.093 ± 0.01 1.43 ± 0.10 0.16 ± 0.01 Anthoxanthum odoratum 18 267.7 ± 10.34 0.703 ± 0.12 1.65 ± 0.09 0.22 ± 0.02 Carmichaelia australis 17 420.6 ± 51.06 0.911 ± 0.18 1.53 ± 0.09 0.14 ± 0.01 Chionochloa conspicua 18 293.8 ± 29.67 0.284 ± 0.06 1.71 ± 0.04 0.17 ± 0.01 Clianthus puniceus 16 190.2 ± 38.24 0.538 ± 0.23 2.82 ± 0.13 0.24 ± 0.03 Cytisus scoparius 17 535.6 ± 60.27 2.240 ± 0.42 2.30 ± 0.06 0.12 ± 0.01 Leptospermum scoparium 17 277.5 ± 21.56 1.021 ± 0.18 1.14 ± 0.09 0.13 ± 0.01 Lupinus arboreus 13 143.1 ± 29.08 0.295 ± 0.10 2.17 ± 0.08 0.08 ± 0.01 Pinus radiata 18 121.1 ± 13.32 0.364 ± 0.08 1.21 ± 0.08 0.16 ± 0.02 Poa colensoi 17 278.1 ± 15.72 0.821 ± 0.23 1.33 ± 0.11 0.15 ± 0.01 Podocarpus totara 18 137.8 ± 16.32 0.335 ± 0.05 1.15 ± 0.04 0.15 ± 0.01 Pseudotsuga menziesii 18 52.8 ± 3.20 0.111 ± 0.01 1.39 ± 0.19 0.13 ± 0.04 Sophora microphylla 18 179.4 ± 28.30 0.275 ± 0.08 1.72 ± 0.15 0.13 ± 0.02 Sophora tetraptera 18 129.7 ± 17.79 0.270 ± 0.07 1.80 ± 0.09 0.22 ± 0.02 Trifolium repens 18 334.7 ± 51.51 1.464 ± 0.36 2.63 ± 0.07 0.18 ± 0.01 Ulex europaeus 18 323.6 ± 29.10 1.347 ± 0.27 2.09 ± 0.05 0.12 ± 0.01 All Fabaceae 135 285.7 ± 18.34 0.936 ± 0.10 2.10 ± 0.05 0.15 ± 0.01 All non-Fabaceae 142 199.4 ± 9.22 0.460 ± 0.05 1.37 ± 0.04 0.16 ± 0.01 All native plants 139 237.2 ± 12.74 0.549 ± 0.06 1.58 ± 0.05 0.16 ± 0.01 All exotic plants 138 245.7 ± 16.56 0.836 ± 0.10 1.83 ± 0.06 0.15 ± 0.01 All native Fabaceae 69 228.4 ± 22.00 0.491 ± 0.08 1.88 ± 0.08 0.18 ± 0.01 All exotic Fabaceae 66 345.7 ± 27.94 1.402 ± 0.18 2.31 ± 0.04 0.13 ± 0.01 All native non-Fabaceae 70 245.9 ± 13.12 0.607 ± 0.08 1.32 ± 0.05 0.15 ± 0.01 All exotic non-Fabaceae 72 154.1 ± 10.58 0.318 ± 0.05 1.42 ± 0.06 0.17 ± 0.01 All ECM plants 53 148.1 ± 15.30 0.489 ± 0.08 1.25 ± 0.08 0.14 ± 0.02 All non-ECM plants 224 263.5 ± 11.91 0.740 ± 0.07 1.82 ± 0.04 0.16 ± 0.00 All AMF plants 228 271.4 ± 11.53 0.787 ± 0.07 1.75 ± 0.04 0.16 ± 0.00 All non-AMF plants 49 101.9 ± 10.54 0.253 ± 0.04 1.49 ± 0.10 0.13 ± 0.02 Scotch broom 17 535.6 ± 60.27 2.240 ± 0.42 2.30 ± 0.06 0.12 ± 0.01 All other exotic Fabaceae 49 279.8 ± 25.55 1.111 ± 0.18 2.32 ± 0.05 0.14 ± 0.01 (excl. broom)

Table 3. Linear mixed-effect model results for C. scoparius % coverage, plant origin, and legume status (top table) and C. scoparius coverage, plant origin, and ectomycorrhizal (ECM) status (below table). Accompanying t-values are compiled in Appendix A2. Non-significant terms only included if part of significant higher level interaction. “.” indicates term dropped during model simplification.

Broom coverage Broom coverage Fabaceae Broom coverage Broom coverage Fabaceae Native × Fabaceae × Native × Native × Fabaceae × Native Shoot mass (g) < 0.0001 0.0365 0.1843 < 0.0001 0.0566 0.0075 < 0.0001 Root mass (g) < 0.0001 0.9374 0.9760 0.0247 0.6540 0.0561 0.0015 Whole plant mass (g) < 0.0001 0.0388 0.2879 < 0.0001 0.0723 0.0092 < 0.0001 Shoot N (%) 0.1327 < 0.0001 0.1181 0.3362 0.9227 0.3750 0.0137 Shoot P (%) . 0.7259 0.3788 . . 0.0340 . Shoot N (%) / P (%) 0.6784 0.0002 0.0072 . 0.0009 0.0497 . Total N / Total P < 0.0001 0.0007 0.0192 < 0.0001 0.0018 0.0023 < 0.0001

Broom coverage Broom coverage ECM Broom coverage Broom coverage Fabacea Native × ECM × Native × Native × ECM × Native Shoot mass (g) < 0.0001 0.3408 0.1987 0.0038 0.0271 0.0463 0.0010 Root mass (g) < 0.0001 0.3515 0.8359 0.0147 . 0.0467 . Whole plant mass (g) < 0.0001 0.4877 0.3006 0.0001 0.0045 0.0888 0.0225 Shoot N (%) 0.1376 0.0247 . 0.0073 . . . Shoot P (%) ...... Shoot N (%) / P (%) 0.6781 0.3451 0.0408 0.0003 0.0056 . . Total N / Total P < 0.0001 0.1624 0.0504 0.0037 0.0005 0.1289 0.0043

Figure 4. Standardised and centred aboveground dry biomass (g) over C. scoparius % coverage for all 16 plant species. Regression lines are shown when P < 0.05.

Figure 5. Shoot % N over C. scoparius % coverage for all 16 plant species. Regression lines are shown when P < 0.05.

Figure 6. N:P (%) ratio over C. scoparius % coverage for all 16 plant species. Regression lines are shown when P < 0.05. Plants above the dashed purple line are putatively P-limited, whereas plants below the dotted green line are putatively N-limited (Koerselman and Meuleman 1996). Plants between both lines can be limited by either N or P or both nutrients.

Figure 7. Standardised and centred total shoot N:P ratio over C. scoparius % coverage for all 16 plant species. Regression lines are shown when P < 0.05.

Table 4. Standardized linear coefficients of soil chemistry and plant traits correlates of plant measurements after model simplification based on drop1 “lme4 (v1.1)” package. Dashes (_) indicate not included. All coefficients obtained from the model after drop1 simplification are reported, regardless if significant. *P < 0.05, **P < 0.01, and ***P < 0.001.

Shoot mass Root mass Whole plant Shoot N Shoot P Shoot N (%) Total N Total P Total N (g) (g) mass (g) (%) (%) / Shoot P (%) (Shoot) (Shoot) / Total P

Broom % coverage 0.447*** 0.287*** 0.312*** 2.418 0.427 2.557 0.217*** 0.968*** 0.024*** Fabaceae 0.928 -0.270 0.865 4.257*** -1.143 3.800** 1.440** 0.730* 1.518** Native 0.194 0.418 0.428 0.051 -0.904 0.298* 0.030 -0.107 0.111*

C (%) ______N (%) ______Olsen P (mg/kg) ______Ca (cmol(+)/kg) 2.59** _ _ _ _ _ 2.677** _ 3.660*** Mg (cmol(+)/kg) -2.087* _ _ _ _ _ -2.075* _ -3.289**

Soilchemistry K (cmol(+)/kg) ______Na (cmol(+)/kg) ______Broom % coverage × Fabaceae 6.606*** 3.816* 5.826*** -2.386 -0.722 -1.774 7.453*** 5.620*** 7.236*** Broom % coverage × Native 2.111 1.977 1.680 -1.699 0.522 -2.980*** 1.096* 2.049 0.952** Fabaceae × Native -0.803* -0.400 -0.907* -1.421 1.360 -1.829 -0.860* -0.547 -0.982** Broom % coverage × Native × Fabaceae -4.990*** -3.140** -3.974*** 2.423* 0.923 0.723 -4.375*** -3.120** -4.871***

Discussion

The results of this study suggest that, compared with uninvaded soil, plants grown in soil with C. scoparius’ legacy have 1) higher above- and belowground biomass and higher total N:P ratios, particularly for native plants, 2) lower root:shoot ratios, 3) no apparent changes in % P, and 4) a mixed response concerning % N, where C. scoparius coverage did not change shoot % N in any tested Fabaceae, yet affected half of the non-Fabaceae. I reject my hypothesis that the soil legacy of C. scoparius has a more positive effect on exotic plants than on natives. Regarding my hypothesis that the soil legacy of C. scoparius will favour members of its own taxonomic family, I conclude that C. scoparius’ soil legacy does not discriminate against its taxonomic family, yet does not necessarily favour non-Fabaceae either. Although some soil chemical traits had a slight correlation with the effect of C. scoparius coverage on plant growth, the biological effect of C. scoparius coverage superceded that of soil chemistry.

General effect of the soil legacy of C. scoparius on biomass

Cytisus scoparius invasion has been associated with an increase in exotic plant species and a decline in native species (Shaben and Myers 2010). For a given plant species, soil obtained from closely related plants has generally been considered to have a more negative effect on plant growth than soil obtained from distantly related plants (Kempel et al. 2018). My results show no strong indication of the soil legacy of C. scoparius having any taxonomic bias, but rather show that C. scoparius favours native plants over exotics. The soil legacy of C. scoparius increased the biomass of all but one of the eight native plant species in my experiment, compared with benefiting only half of the tested exotic species. Although the mean biomass of invasive Fabaceae outweighed that of native Fabaceae, once normalized, the effect of C. scoparius’ soil legacy did not discriminate between native and exotic Fabaceae in terms of biomass, nor between Fabaceae and non- Fabaceae.

It is likely that unaccounted factors contributed to the shrub’s dominance in-field, such as the rapid seedling growth rate of C. scoparius (Grotkopp and Rejmánek 2007) and the tendency of C. scoparius to reduce the availability of light and soil water needed by other plants (Watt et al. 2003b, Wearne and Morgan 2004). We found in a field experiment that plant growth next to live C. scoparius was most beneficial for exotic legumes compared to native legumes (Allen et al. (2020); Appendix E), yet especially when the rhizosphere of C. scoparius was allowed to influence plant growth. I can therefore speculate that the belowground association of C. scoparius with soil organisms (Grove et al. 2017) is another factor contributing to the dominance of C. scoparius in-field and may underlie the results of my soil legacy experiment. One possible reason as to why field-based studies on the

30 performance of native NZ plants showed the negative responses to C. scoparius invasion (e.g. Allen et al. 1995; Watt et al. 2003) is that native plants still need to deal with co-evolved antagonists in situ (e.g. specialist pathogens). A possible consideration would be that beneficial effects of mutualists associating with C. scoparius may have shifted to the native plants, thereby counterbalancing the antagonists.

Plant root:shoot ratios are likely to decline when conditions for growth improve as a result of increased soil moisture or nutrient availability (Wilson 1988). As moisture availability was uniform for all plants due to growing in greenhouse conditions, seeing a decline in root:shoot ratio, which was most apparent for two native species, suggests that C. scoparius increases soil nutrient availability.

Plant specific effects of C. scoparius soil legacy on biomass

It was expected that shoot biomass did not increase with C. scoparius coverage for Pseudotsuga menziesii as negative effects of C. scoparius on P. menziesii, driven largely by reduced ectomycorrhizal fungal colonization, have been documented (Grove et al. 2012). As Pinus radiata is also primarily ectomycorrhizal (Teste et al. 2020), finding no increase in biomass with C. scoparius coverage was also expected and could broadly be attributed to the need for ectomycorrhizal seedlings to have appropriate mycorrhizal inoculum for effective growth (Lekberg and Koide 2005, Dickie et al. 2012).

It was less expected to find an increase in the biomass of native ectomycorrhizal Leptospermum scoparium (mānuka), which likewise had to contest with a probable reduction in available ectomycorrhizal inoculum. However, among native New Zealand plants which form ectomycorrhizal associations, L. scoparium is known to associate with both arbuscular mycorrhizal and ectomycorrhizal fungi (Davis et al. 2013). Being dual-mycorrhizal, L. scoparium could have opted into associating with arbuscular mycorrhizal fungi (Teste et al. 2020), which could partially explain why L. scoparium was not hindered in its growth as C. scoparius coverage increased.

Effect of C. scoparius soil legacy on % N and % P

Based on nutrient limitation indicators proposed by Koerselman and Meuleman (1996), where plant N:P ratios >16 indicate putative P limitation and plant N:P ratios <14 indicate putative N limitation, it can be observed that exotic Fabaceae are prone to being P-limited whereas all other plants, particularly the grasses and native non-Fabaceae, are N-limited. Sophora microphylla was here a unique case in that increased C. scoparius coverage seemed to transition the plant from a state of putative P limitation to putative N limitation, however this transition was not observed for other native Fabaceae.

The direction of changes in shoot % N did not necessarily correlate with corresponding changes in shoot biomass. As live C. scoparius presence has been associated with an increase in the cover of exotic Anthoxanthum odoratum (Carter et al. 2019c), seeing an increase in the aboveground biomass of A. odoratum with C. scoparius coverage was expected, yet not alongside a decrease in aboveground % N. On the other hand, the aboveground biomass of both Pinus radiata and Agrostis capillaris remained unchanged with C. scoparius coverage, even though both plants showed increases in shoot % N over C. scoparius coverage. Although Watt et al. (2003b) had shown that juvenile Pinus radiata grew markedly less alongside live C. scoparius, the same authors suggested that C. scoparius may in the long-term enhance the growth of Pinus radiata in N-deficient sites, provided that water stress did not impair nutrient uptake (Nambiar and Sands 1993). Observing an increase in the shoot % N of Pinus radiata without seeing any change in shoot biomass over C. scoparius coverage could be attributed to pine in general having specific growth spurts even if unrestricted by available N (Fagerström and Lohm 1977). Regarding A. odoratum’s decrease in % N despite an increase in aboveground biomass, A. odoratum is known to be a relatively fast growing grass which accumulates large quantities of soil N compared with slower growing grasses (Weigelt et al. 2005). The simplest explanation to A. odoratum’s decrease in % N is that the grass is responding to some other limiting resource, possibly soil C.

Effect of C. scoparius coverage on soil chemistry

Although C. scoparius is known to increase soil organic C (Fogarty and Facelli 1999), the availability of soil P (Dewar et al. 2006) (although see Shaben and Myers (2010)), as well as soil N (Watt et al. 2003b), there was no correlation observed between C. scoparius coverage and soil chemistry in my data. Broadbent et al. (2017), who used the same data on soil properties yet separate data on C. scoparius coverage, also reported that C. scoparius coverage frequently led to no significant difference in soil properties. Caldwell (2006) observed that soil under C. scoparius had a significantly higher C:P ratio than soil from an adjacent area uninvaded by C. scoparius, yet I found no significantly different C:P ratio in my data. Finding no correlation between C. scoparius coverage and soil chemical attributes was least expected for soil N, as soil N has often been documented to increase following a C. scoparius invasion (Haubensak and Parker 2004, Caldwell 2006, Grove et al. 2015), although no changes in soil N caused by C. scoparius have also been reported (Shaben and Myers 2010, Carter et al. 2019c). Change in soil N under C. scoparius has been described as site-specific (Slesak et al. 2016), which could be the case in my data.

It has been suggested that the competitive ability of C. scoparius correlates with nutrient availability (Fogarty and Facelli 1999). As an explanation as to why no correlation between soil nutrients and C. scoparius coverage was found in this experiment, it could be proposed that C. scoparius (or other plants in the field) had already taken up any additional available nutrients which C. scoparius may

32 have imparted on the soil or, more simply, that no significant changes in soil chemical composition were caused by C. scoparius.

Experimental design considerations A common issue with soil legacy studies is that it is difficult to separate the chemical, physical and biological properties in a given soil which can each play differing roles in determining plant growth (Van der Putten et al. 2013). Biological properties can be particularly underrepresented: certain plants exert a disproportionately great influence on soil biota (Weir et al. 2004), even with a small biomass relative to other plants (Peltzer et al. 2009). In contrast, the effect of certain biochemical plant properties might have been overrepresented: C. scoparius has been considered as putatively allelopathic in a study by Grove et al. (2012), who amended soils from four different sites with sucrose, activated C and C. scoparius litter and examined the effect of C. scoparius’ soil legacy on a the growth of a single pine species (Pseudotsuga menziesii). Allelopathy has been a contentious topic since its conception (Williamson 1990) (but see Bais et al. (2003), Pardo-Muras et al. (2020)) and my observed overall increase in plant biomass over C. scoparius coverage gives no indication of allelopathy. The effect of C. scoparius on soil N has already been deemed site-specific (Slesak et al. 2016) and it may be proposed that effects of C. scoparius’ soil legacy attributed to allelopathic compounds might in fact be due to the shrub’s association with certain site-specific microorganisms.

Although no obvious correlation has been observed in my dataset (Appendix A6), soil under a C. scoparius invasion in New Zealand has been found to have significantly higher available P than uninvaded native soil (Dewar et al. 2006). There is some uncertainty as to whether C. scoparius caused an increase in available P or vice-versa, as superphosphate fertilizers with long-lasting effects have been applied to large areas of New Zealand by aerial means (During 1972, Sharpley and Syers 1979, Will et al. 1985). Caution is needed when drawing conclusions related to soil nutrients, as these are known to vary between seasons (Powers 1990, Gilliam et al. 2001), yet my soil nutrient data (used by Broadbent et al. (2017)) should still reliably reflect C. scoparius presence as no recent change in C. scoparius invasion to the Molesworth plots was observed.

It is notable that native grasses grown in competition with C. scoparius have been shown to inhibit C. scoparius development (Harrington 2011, Lang et al. 2017), and the Molesworth field site used in this experiment does have near mono-dominant patches of C. scoparius in close proximity to scarcely disturbed grassland. I cannot rule out the possibility that the soil legacy of the native grasses decreased plant biomass (as opposed to C. scoparius’ soil legacy being solely responsible for the observed trends). A future research direction would be to include sterilized soil treatments when performing a similar experiment to account for the putative effects of such native grasses or other biotic variables.

Conclusions and applications

Although in-field observations imply the opposite, the soil legacy of C. scoparius unexpectedly favours the growth of native New Zealand plants over its own taxonomic family. Given that C. scoparius’ predominantly positive soil legacy effect can only be partly attributed to soil chemical traits, microbial effects could very well play an important role in the invasion success of C. scoparius (Haubensak and Parker 2004). Compared with classic soil legacy studies, much less is known on the importance of soil legacies involving possible co-invading belowground mutualists (Nuñez and Dickie 2014). My next chapter will delve deeper into the effect of C. scoparius on surrounding microbial composition, specifically that of soil fungi.

Chapter 3: The response of fungal communities to Cytisus scoparius invasion

Abstract

Homogenization caused by introduced plants typically results in decreases in the species richness of associating organisms, at least at the local scale. Although studies on the response of communities to a plant invasion have often taken an aboveground focus, less is known regarding the belowground consequences of a plant invasion, despite soil communities being integral to conservation efforts and having a long-lasting effect on plant community composition. To study the effect on an invasive nitrogen-fixing plant on soil fungal communities, I examined metabarcoded environmental DNA data from soil across a natural Cytisus scoparius density gradient. I categorised fungal operational taxonomic units into specific functional guilds such as saprotrophs and plant pathogens, which allowed a more thorough examination of how C. scoparius affected fungal communities. Although I found that certain fungal groups became more homogeneous under higher densities of C. scoparius (likely caused by increased plant homogeneity), my results showed that C. scoparius invasion increased average fungal diversity at the point-scale per plot (likely caused by increased soil productivity). A greater proportion of unique fungal taxonomic units were found near C. scoparius as opposed to uninvaded grassland and soil under C. scoparius had a higher richness of saprotrophs, plant pathogens and arbuscular mycorrhizal fungi. Despite leading to a lower plant diversity, it is possible that changes in soil properties induced by C. scoparius, such as increased N deposition and water retention, enabled a higher richness of fungi to live alongside the invasive shrub. My results indicate that coalescence between previously separated fungal communities may have occurred due to C. scoparius. Apart from C. scoparius having a definite effect on soil fungal communities, it is possible that the soil fungal communities themselves might contribute to the shrub’s invasiveness.

Keywords

Arbuscular mycorrhizal fungi, diversity, environmental DNA, soil communities, metabarcoding

Introduction

Soil biodiversity patterns are moulded by a wide range of (a)biotic factors which operate at different spatial scales (Ettema and Wardle 2002). At the smallest spatial scale (micrometre to millimetre), distribution patterns of soil biota are in part affected by rooting patterns in plants and microscale soil heterogeneity (Bardgett and Van Der Putten 2014). At a fine-scale (millimetre to centimetres), spatial patterns in microbial communities are influenced to some extent by root exudates (Broeckling et al. 2008), which attract microbial symbionts to roots, including rhizobia, mycorrhizal fungi (Badri and Vivanco 2009), and soil pathogens (Mendes et al. 2011). Chemical and physical soil properties (e.g., soil nutrient availability and soil water) alongside the identity of dominant plants determine spatial patterns of soil biota at the local scale (centimetres to metres) (Wardle 2013). At even larger regional and continental scales, which can range from metres to hundreds of kilometres, dynamics such as topography, climate and continental isolation play a more important role (Fierer and Jackson 2006).

Plant-mediated changes to soil microbial composition at the local scale (centimetres to metres) are virtually omnipresent in natural environments, as soil micro-organisms form symbiotic interactions with most plants (Smith and Read 2010) and these plants can have a profound influence on the structure of surrounding microbial communities (Tedersoo et al. 2016, Kivlin et al. 2018). Most research on a plant’s effect on soil communities has taken a plant-soil feedback approach, measuring plant growth responses but not identifying species of microbes (Bever et al. 2010). In spite of an increase in plant-soil feedback studies, relatively little is known about how plants modify soil microbial community composition (Eisenhauer et al. 2010, Maul and Drinkwater 2010). Even at low biomass, some plants are capable of “punching above their weight” in terms of modifying soil properties (Peltzer et al. 2009) and invasive plants can cause changes in soil microbial communities with long-lasting consequences to ecosystem function (Nuñez and Dickie 2014). The abundance, activity and composition of soil microbial communities have long been known to vary with different plant species (Bever et al. 1997), yet further than just passively affecting microbial communities in their surrounding soil, plants have been known to actively cultivate their own associating micro-organisms (Broz et al. 2007, Broeckling et al. 2008). Such affiliations between plants and belowground mutualists can be very specific (Stefanowicz et al. 2019), with the spread of certain plants having been linked with a single species of associating micro-organism (Hayward et al. 2015). Both the plant and the plant’s mutualists have become invasive in certain cases (Marler et al. 1999, Simberloff and Von Holle 1999, Richardson et al. 2000, Callaway et al. 2001), with profound consequences on ecosystem development (Dickie et al. 2019).

Given the numerous spatial scales at which soil biota associated with a plant invasion can be studied, there is limited information on how an invasive plant can change belowground microbial diversity and heterogeneity across different spatial scales. Although changes in soil community composition have been analysed in successional studies (Van der Putten et al. 1993, Dickie et al. 2019) and across environmental and latitudinal gradients (Sharp et al. 2014, Tedersoo et al. 2014, Lu et al. 2018), the belowground consequences of invasive plants on soil biota remain less known (Van der Putten et al. 2007b).

There is generally a positive relationship between the biodiversity of groups of directly or indirectly interacting organisms (Gaston 2000, Scherber et al. 2010, Peng et al. 2019), particularly in terms of beta diversity. As such, it is expected that increased plant diversity correlates with increased belowground fungal diversity, as has been shown for ectomycorrhizal fungi (Dickie 2007, Lang et al. 2011, Gao et al. 2013). The richness of arbuscular mycorrhizal fungi (AMF) is also positively associated with plant species richness (Vogelsang et al. 2006, Hiiesalu et al. 2014), although no association between plant richness and AMF richness has likewise been recorded (Öpik et al. 2008, Lekberg et al. 2013). It is known that a plant invasion can actually increase both the abundance and diversity of AMF compared to native uninvaded soil (Lekberg et al. 2013).

While decreases in local diversity caused by an invasive plant are typical (Hejda et al. 2009, Powell et al. 2011, Lishawa et al. 2019), plant-induced increases in local diversity have been related to improved ecosystem productivity (Liao et al. 2008, Ehrenfeld 2010). Nutrient cycling commonly underlies ecosystem productivity (Ehrenfeld 2003, Daryanto et al. 2019) and may be enhanced by interactions between plants and associated AMF (Saia et al. 2020b) and/or N-fixing Rhizobia. Although a meta-analysis on the effects of both alien N-fixing plants and alien non-N-fixing plants has shown similar degrees of impact on native plant communities (Vilà et al. 2011), this pattern does not address belowground soil biota. Changes in vegetation caused by N-fixers can have a very long-lasting effect on N availability and nutrient cycling (Hu et al. 2001), which may in turn affect the composition of soil biota. Simulated N deposition has been correlated with decreases in AMF root colonization and spore production (Van Diepen et al. 2011), alongside decreases in AMF diversity (Wang et al. 2011, Lin et al. 2012), ectomycorrhizal fungi diversity (Wright et al. 2009) and general soil fungal diversity (Edwards et al. 2011, Paungfoo-Lonhienne et al. 2015, Zhou et al. 2016). N deposition also has a potential to negatively impact on C cycling in soil and may promote fungi which have pathogenic traits (Paungfoo-Lonhienne et al. 2015). Although decreases in fungal diversity are most common following N deposition, outcomes may vary according to the studied system as N fertilizer has been shown to both increase (Klaubauf et al. 2010) and decrease (Edwards et al. 2011) the relative abundance of the fungal phylum Ascomycota (the most common fungal phylum in both studies).

Most studies which focus on the effect of invasive plants on soil fungal communities use experimental approaches rather than undertaking a natural survey. For example, Gornish et al. (2016) simulated different levels of plant invasion by seeding plots of Taeniatherum caputmedusae, as opposed to sampling across a natural invasion gradient. Burke et al. (2019) and Anthony et al. (2019), who both studied the effect of invasive Alliaria petiolate on fungal communities, used exclosure and control plots and an eradication-based experimental design, respectively.

How soil communities are identified and categorised may also be an important factors. Gaggini et al. (2018) found that invasive Impatiens glandulifera increased soil fungal community diversity. In a study on how three exotic grasses added to experimentally grown native monocultures modified soil microbial composition, Gibbons et al. (2017) found no significant changes in fungal alpha and beta diversity caused by any of their studied invasive grasses. However in the same study, when fungal operational taxonomic units (OTUs) were assigned to functional guilds, two exotic grasses caused a decrease in fungal OTUs assigned as symbionts and one exotic grass caused an increase in fungal OTUs assigned as pathogens, whereas none of the three tested grasses caused a change in the richness of saprotrophic fungi (i.e., decomposers).

At a broader scale, fungal species identified as saprotrophs are known to have an increased abundance in native forests compared to forests invaded by Ligustrum lucidum (Fernandez et al. 2017) and fungal species diversity has generally been found to decrease with increased density of invasive plants such as Centaurea maculosa (Broz et al. 2007), Ageratina adenophora (Balami et al. 2017) and Robinia pseudoacacia (Liu et al. 2018), although an increase in fungal richness was observed for Alliaria petiolata (Anthony et al. 2017). All the above studies either use artificially created monocultures or a direct “control vs. effect” experimental design instead of sampling across an in-situ gradient of plant invasion. Moreover, none of the studied plants in these studies, apart from Robinia pseudoacacia (Liu et al. 2018), are N-fixers.

There is little known regarding the effect of my species of interest, C. scoparius, on microbial communities (although see Johnston et al. 1995). Compared with uninvaded soil, C. scoparius increases microbial biomass N, microbial biomass P and microbial biomass C (Dewar et al. 2006), hence a case can be made that the proportional composition of certain fungal taxa might increase following a C. scoparius invasion.

I chose to focus on soil fungi as they have a better “species concept” than bacteria, in part as functional genes are less frequently horizontally transferred between fungal species than in bacteria (Klingmüller et al. 1990). Soil fungi have also been observed as being more spatially heterogeneously distributed in soil compared to bacteria (Manter et al. 2010), and are more responsive to the identity of specific invasive plants (Maron et al. 2011, Xiao et al. 2014, Bahram et al. 2020). With regard to plant pathogens, most plant pathogen OTUs across different landscapes

38 in New Zealand have been identified as fungi as opposed to bacteria or oomycetes (Makiola et al. 2019a). Studying fungal OTUs is hence likely to give an accurate depiction of how an invasive plant modifies soil pathogen communities, especially as fungal pathogens are more intimately associated with living plants compared to most bacteria (Webster and Weber 2007, Bahram et al. 2020). With regard to fungal saprotrophs, these have also been known to outnumber bacteria within the decomposer community (Maraun and Scheu 1996).

Using spatially explicit data on fungi collected across the density gradient of an invasive plant, my aim was to survey how soil fungal communities responded to a C. scoparius invasion. Seeing near- monocultures of invasive plants is an obvious indicator of decreased plant diversity and increased plant homogeneity (Waterhouse 1988), which should accompany a decrease in the diversity of directly or indirectly associating organisms (Gaston 2000, Scherber et al. 2010). Anything contributing to N input in an ecosystem, such as C. scoparius’ association with N-fixing rhizobia, is also likely to increase homogenization (Olden 2006). Here I look at three levels of diversity. I use “gamma diversity” as the number of fungal species found in a 20 × 20 m plot, and alpha diversity as the number of fungal species found in a soil core. I measure beta diversity (fungal community heterogeneity) as the ratio between plot level gamma diversity and soil core level alpha diversity (Whittaker 1970).

I hypothesised that:

• C. scoparius invasion will result in a decrease in soil fungal gamma diversity and alpha diversity.

• C. scoparius invasion will result in a decrease in soil fungal heterogeneity (beta diversity) within 20 × 20 m plots.

• Given the mostly positive soil legacy of C. scoparius observed in my soil legacy experiment (Chapter 2), that soil invaded by C. scoparius will contain fewer antagonists.

• Given that C. scoparius has been known to increase microbial biomass, that soil invaded by C. scoparius will undergo shifts in the proportional abundance of certain taxa and/or functional groups.

In addition to the hypotheses, I also investigated how C. scoparius might affect fungal community composition and OTU occupancy across plots, enabling a more comprehensive view of belowground changes which correlate with C. scoparius coverage. Metabarcoding is well suited to deal with the diversity and identity of soil fungi in novel ecosystems (Dickie and St John 2016) and has recently been successfully applied to study large-scale patterns in the distribution of soil biota (Makiola et al. 2019a). To test these hypotheses, I examine metabarcoded eDNA data from

432 georeferenced soil cores extracted from 18 plots (24 soil extracts per plot) across a natural C. scoparius density gradient. I categorised fungal OTUs into specific functional guilds such as saprotrophs and antagonists, which allowed a more thorough examination of C. scoparius’ effect on co-occurring fungi (Zanne et al. 2019).

Methods

Study site and natural experiment

I undertook field sampling from 14 February 2017 to 17 April 2017. For each of the 18 permanent vegetation plots, I extracted 24 individual georeferenced soil cores (Figure 2), for a total of 432 spatially explicit soil samples. To reliably pinpoint areas for soil extraction, I subdivided each plot into 5 × 5 m subplots by laying out measuring tapes at 5 m intervals. All extractions were taken at their precise planned location and no sampling deflection (due to immovable objects, e.g., tree trunks) was necessary. I kept a catalogue of natural occurrences which might affect the composition of extracted soil (e.g., soil was dug up from within an active ant-nest and beneath a decomposing hare carcass) and I recorded the following C. scoparius coverage measurements at each point of soil extraction: 1) distance to closest immature C. scoparius, 2) distance to closest mature C. scoparius (determined by the presence of flowers or seedpods), 3) height of the tallest C. scoparius within 1 m radius, and 4) a C. scoparius density estimate within 1 m radius, which I refer to as “broom coverage”.

Prior to soil extraction, metal trowels were manually scrubbed and then sterilized in a 10% v/v bleach solution (8% sodium hypochlorite in undiluted bleach) for >10 min (Prince and Andrus 1992) before being rinsed in water. I dug up soil samples vertically to a depth of 150-200 mm, each soil sample weighing 200–250 g. Litter and leaf matter, typically forming the top 10-20 mm of a soil sample, was discarded. Once dug up, soil samples were sealed in individual, clean, zip-lock plastic bags and transported in insulated ice chests prior to being stored at 4°C until subsequent processing. The time between field extraction and refrigeration at 4°C was restricted to a maximum of 60 hours and the soil samples were left refrigerated until further processing for a maximum of 14 days. The collected samples were stored in a fridge temporarily, as opposed to a freezer, as the

41 soil would otherwise be subject to unwanted freeze-thaw cycles, which disturbs microbial communities (Pesaro et al. 2003). Each refrigerated soil sample was broken up manually and spread out evenly on clean paper. Using bleached forceps and spatulas, I obtained a ~10 g mixed soil sample by systematically extracting 10 × ~1 g subsamples from across the initial sample. The mixed subsampled soil did not contain any roots more than 5 mm in width or stones larger than 5 mm in diameter and obvious insects (e.g., ants, larvae) were avoided. I kept the processed soils frozen at -18°C until DNA extraction.

Wet-lab processing

Both the kit used for soil DNA extraction and the chosen fungal primers were recommended by Lear et al. (2018). I performed DNA extraction on the 432 soil cores using DNeasy PowerSoil® HTP 96 Kits (Quiagen), according to the manufacturer’s instructions and loading the maximum amount of recommended soil for DNA extraction (250 mg). As part of the PowerSoil® protocol, mechanical lysis of the soil samples was performed using a Spex® Sample Prep 1600 MiniG. Based on amplification protocols outlined by the Earth Microbiome Project (Gilbert et al. 2014) (http://press.igsb.anl.gov/ earthmicrobiome/protocols-and-standards/its/), I assembled two single-indexed DNA libraries from the 432 soil extracts using the fITS7 general fungal primer (5′- GTG ART CAT CGA ATC TTT G -3′) (Ihrmark et al. 2012) and the ITS4 reverse primer (5’- TCC GCT TAT TGA TAT GC -3’) (White et al. 1990). The ITS4 reverse primer was designed with both Illumina adapter sequences and index sequences (Caporaso et al. 2011), permitting future identification of the sequenced amplicons. I ordered the fITS7 primer (along with its Illumina adapter sequence) from Integrated DNA Technologies (Purification method: Standard Desalting). The Illumina adapter for fITS7 was 5’- AAT GAT ACG GCG ACC GAG ATC TAC AC -3′ and the Illumina adapter for ITS4 was 5’- CAA GCA GAA GAC GGC ATA CGA GAT -3′.

Using an Eppendorf vapoprotect Mastercycler®, PCR amplifications were performed in a 25 µL mixture volume containing 0.2 μL FastStart™ DNA polymerase (Merck), 0.5 μL dNTP mixture

(10 mM each), 2.5 μL PCR buffer (with 20 mM MgCl2, sourced from Merck), 2 μL 2.5 μM of each forward and reverse primer, 1.25 μL 10 μM molecular grade Bovine Serum Albumin, 1 μL 10× diluted DNA template and 15.55 μL filtered deionized water (obtained via a Milli-Q® water purification system and filtered through a Biopak® Polisher). I used Bovine Serum Albumin to reduce the effect of PCR inhibitors derived from soil (Jiang et al. 2005) and assembled all PCR reagents prior to adding the DNA template in a dedicated UV light irradiated chamber with a dedicated set of micropipettes. PCRs were carried out under the following conditions: a denaturation step of 5 min at 94°C, followed by 35 cycles of 30 s at 94°C, 30 s at 57°C and 30 s at 72°C, with a final step at 72°C for 7 min (and held at 4°C). All PCRs were carried out in duplicate along with positive and negative controls. To confirm amplification, I performed agarose gel

42 electrophoresis on the PCR product, stained with RedSafeTM (iNtRON) and using a 1% agarose gel. No PCR products were observed in negative controls and samples which showed poor amplification were rerun. The 10× dilution of the DNA templates (with filtered deionized water) was undertaken after initially performing PCRs with 1 μL of the undiluted PowerSoil® DNA elution, which yielded less PCR product according to gel runs, possibly due to PCR inhibition caused by soil components such as tannins (Kreader 1996). As the PCRs were performed in duplicate, 20 μL of each duplicate PCR product were combined prior to normalization.

Following manufacturer’s instructions, I used SequalPrepTM Normalization Plates (ThermoFisher Scientific) to both clean the obtained amplicons and to normalize their concentration relative to each other. Twenty-five μL of mixed duplicate PCR product (25 μL being the recommended maximum) underwent normalization as part of the SequalPrepTM protocol. I eluted my normalized PCR product with 12 μL elution buffer (instead of 20 μL recommended by the manufacturer) due to a previous unsuccessful sequencing library submission that might possibly have been caused by too low a concentration of PCR amplicons. The resulting concentration of normalized samples was 2.14 ng/μL. I prepared two sequencing libraries; for each library, 12 μL of normalized indexed PCR amplicons were pooled together, vortexed and 110 μL each of the two resulting mixtures were sent to Massey Genome Service, New Zealand, to undergo Illumina MiSeqTM (2×250 base PE v2) (Caporaso et al. 2012).

Bioinformatics and statistical analysis

I merged forward and reverse Illumina reads using a 32-bit version of USEARCH v11.0.667 (Edgar 2010). I removed any sequences with less than 200bp or which had more than one expected error using VSEARCH 2.10.4 (Rognes et al. 2016). In order to increase the qualitative nature of the sequencing reads and to account for PCR and sequencing artefacts (Leray and Knowlton 2017) and singletons (Dickie 2010), any sequences occurring either once or twice were removed, while the remaining sequences were clustered to 97% similarity threshold. Although both USEARCH and VSEARCH could have been used to filter sequences, the 64-bit version of VSEARCH is open- source and was therefore chosen. OTUs were matched using BLAST v2.5.0+ (Altschul et al. 1997) against the UNITE public database (accessed July 2019) (Nilsson et al. 2018). I removed all recorded OTUs which were not within the kingdom Fungi and all OTUs which had a <200 bp match to any known species. Extraction blanks, and positive and negative controls were checked for contamination and OTUs which were found within my negative controls (0.34% of all OTUs) were also deleted. In order to further limit the effects of PCR and sequencing artefacts (Vesty et al. 2017), I excluded low abundance OTUs by setting OTU occurrence to 0 if any OTU occurred less than 3 times in any sample. In the case where the same soil extract was sent to be sequenced twice

(due to an insufficient amount of amplification observed via gel runs), the sample with the lowest number of reads was deleted along with any sample which had <1000 reads (0.23% of samples).

In order to classify fungal OTUs according to functional guild, OTUs were matched against the FUNGuild database (Nguyen et al. 2016). FUNGuild has shown previous effective use in identifying putative soil symbiotrophs, saprotrophs and pathogens (e.g., Gibbons et al. (2017), Lu et al. (2018)). Some fungi had multiple ecosystem functions and as many OTU classifications listed by FUNGuild overlapped with each other (e.g., an OTU could be described as both a saprotroph and a pathogen), I created two databases, one in which each OTU was strictly classified according to a single functional guild, and one in which OTUs were loosely classified and could overlap between guilds. As an example, a ‘strictly’ classified OTU would only be described by FUNGuild as being a saprotroph or a pathogen (and never both), whereas a ‘loosely’ classified OTU could be classified as both a saprotroph and a pathogen. Regarding putative pathogens in both ‘strictly’ and ‘loosely’ categorised databases, further subdivisions were created for each database: 1) a group in which all known antagonist OTUs were compiled, 2) a group specifically for antagonists of fungi and 3) a group specifically for plant antagonists.

I used R version 3.5.0 (Team 2013) for creating graphs and conducting analyses. All diversity estimates were based on evenly rarefied OTU matrices and the subsample size for rarefying my community was set to the minimum number of sequences in any sample. When examining the composition of my samples according to fungal taxon and functional guild, I first generated randomly rarefied community versions of my dataset via the rrarefy function in “vegan” (Oksanen et al. 2013). This process was iterated 250 times before calculating average alpha, beta and gamma diversities and proportional abundances for each fungal taxon and functional guild across all iterations. The rrarefy function was here appropriate as random rarefaction is performed without replacement so that the variance of community metrics is not related to the size of the sample. When calculating beta (β) diversity for each plot I used the below formula for true beta diversity from Whittaker (1970), where gamma (γ) is a the total species diversity in any given plot and alpha (α) is the average diversity across all soil cores (min. 23) per plot.

True beta diversity has been used in large-scale fungal studies (e.g., Kivlin et al. 2011). To quantify the effects of C. scoparius coverage on fungal alpha diversity and proportional abundance at the level of soil cores, I used linear mixed-effect models via the R package “lme4 (v1.121)” (Bates et al. 2014), setting sampling plot as a random effect. The mixed models break down the variance into inter- and intraplot components and thus improve the true structure of the randomness present in the data (Millar and Anderson 2004).

Figure 1. Picture of the Molesworth field site taken in March 2017 and prominently featuring a C. scoparius (Scotch broom) invasion.

Figure 2. Overview of soil core collection from each of 18 field plots (total soil cores = 432).

Results

In total, 5263 fungal OTUs were identified from 431 soil extractions across 18 plots, the three most dominant fungal phyla and subphyla being Ascomycota (3078 OTUs; 58.5%), Basidiomycota (1606 OTUs; 30.5%) and Mortierellomycotina (152 OTUs; 2.9%) (Appendix B1). Eight out of the ten most abundant OTUs belonged to Ascomycota. A few OTUs dominated the community, with around half (50.9%) of reads belonging to the 60 most abundant OTUs (1.1% of all OTUs). Rare OTUs were common, with 57.2% of OTUs (n = 3008) occurring only between three to ten times across all samples (three being the decided minimum for any soil sample). In functional assignments, 29% of OTUs were identified as ‘strict’ saprotrophs, 13.9% as ‘strict’ symbiotrophs and 8.1% as ‘strict’ antagonists; but most OTUs were classified as having multiple ecosystem functions.

Overall gamma (γ) diversity

There was no significant change in gamma diversity over C. scoparius coverage for all fungal OTUs, nor for any individual fungal taxon or functional guild apart from Glomeromycotina (R2 = 0.59; P < 0.0001) and Mucoromycotina (R2 = 0.64; P = 0.0001; Appendix B2), which both showed increases in gamma diversity over C. scoparius coverage (Figure 3). Although no response in gamma diversity over C. scoparius coverage was observed for Ascomycota, the most abundant taxon across all plots, the gamma diversity of Ascomycota relative to the gamma diversity of all fungal OTUs decreased over C. scoparius coverage (R2 = 0.37; P = 0.0075; Appendix B2).

Overall alpha (α) diversity

For all fungal OTUs, there was a significant positive correlation of alpha diversity (level of soil cores) and C. scoparius coverage (t = 2.289; P = 0.0221) as well as a significant correlation for alpha diversity and log-transformed distance from the extracted soil core to the base of the closest C. scoparius (t = -3.025; P = 0.0025) (Table 1; accompanying t values in Appendix B4). There was no significant correlation between the alpha diversity of Ascomycota and C. scoparius coverage, yet a highly significant correlation between the alpha diversity of all OTUs excluding Ascomycota and C. scoparius coverage (t = 4.662; P < 0.0001) (Appendix B5).

For average alpha diversity of all fungi at the plot-level, there was a significant increase over C. scoparius % coverage and an associated significant decrease over log-transformed distance from the extracted soil core to the closest mature C. scoparius (Figure 4). At the plot-level, average Ascomycota alpha diversity was not correlated with C. scoparius coverage, however all fungal OTUs excluding Ascomycota, and most notably Basidiomycota and Glomeromycotina, increased in alpha diversity over C. scoparius coverage (Figure 5). A positive correlation was likewise found

46 between C. scoparius coverage and Chytridiomycotina alpha diversity (R2 = 0.41; P = 0.0040), as well as the alpha diversity of ‘strict’ saprotrophs (Figure 5).

The results were generally robust to the method of measuring C. scoparius cover and the definition of functional guilds. Average alpha diversity of all fungi at the plot-level had a similar response to mature C. scoparius (R2 = 0.25; P = 0.0361) compared with immature C. scoparius (R2 = 0.22; P = 0.0476) (Appendix B6). When substituting C. scoparius % coverage with the log-transformed distance from the extracted soil core to the closest mature C. scoparius (mm), there were no qualitative changes in any results related to alpha diversity (Appendix B7). Despite the overall increase in average alpha diversity which correlates with C. scoparius coverage, the plot with the highest gamma diversity (MW19) had no C. scoparius (Appendix B2).

Relative changes in average alpha (α) diversity at the level of plots

The ratio of mean Basidiomycota alpha diversity (per plot) compared to the mean alpha diversity of all fungi (per plot) increased over C. scoparius coverage (R2 = 0.23; P = 0.0462). Similar relative increases in mean alpha diversity (per plot) over C. scoparius coverage could be observed for Glomeromycotina (R2 = 0.50; P = 0.0010) and Chytridiomycotina (R2 = 0.41; P = 0.0049) yet not for other fungal taxa nor for saprotrophs or symbiotrophs, although there was a non-significant increasing trend observed for ‘strict’ saprotrophs (R2 = 0.20; P = 0.0660).

Overall beta (β) diversity

For all fungal OTUs, there was no significant change in beta diversity over C. scoparius coverage. When looking at individual fungal taxa and functional guilds, plots with higher C. scoparius coverage showed decreased Glomeromycotina and Basidiomycota beta diversity, whereas the beta diversity of Mucoromycotina increased with C. scoparius coverage (Figure 6). There was no qualitative change in results when substituting C. scoparius coverage with the log-transformed distance from the extracted soil core to the closest mature C. scoparius (Appendix B9). The beta diversity of Chytridiomycotina (R2 = 0.32; P = 0.0180) and saprotrophs (R2 = 0.26; P = 0.0297) both decreased in plots with higher C. scoparius coverage, although in both cases the method of measuring C. scoparius determined whether or not a significant change could be observed.

While ‘strict’ ectomycorrhizal fungi showed no change in beta diversity over C. scoparius coverage, an increase in beta diversity over C. scoparius coverage was found for ‘loose’ ectomycorrhizal fungi (i.e., ectomycorrhizal fungi which shared one or several different functional traits) (R2 = 0.31; P = 0.0169).

Plant pathogens

Gamma. No change in gamma diversity was observed for ‘strict’ plant pathogens over C. scoparius coverage. When adapting the ‘loose’ definition of plant pathogens, the gamma diversity of ‘loose’ plant pathogens relative to the gamma diversity of all fungal OTUs decreased over C. scoparius coverage (R2 = 0.35; P = 0.0100).

Alpha. There was a significant positive correlation of alpha diversity (level of soil cores) and C. scoparius coverage for ‘strict’ plant pathogens (t = 3.014; P = 0.0026) (Table 1). Increases in C. scoparius coverage correlated with increased mean alpha diversity of ‘strict’ plant pathogens (Figure 5). The ratio of mean ‘strict’ plant pathogen alpha diversity (per plot) compared to the mean alpha diversity of all fungi (per plot) increased over C. scoparius coverage (R2 = 0.29; P = 0.0213).

Beta. Beta diversity of ‘strict’ plant pathogens decreased over C. scoparius coverage (R2 = 0.50; P = 0.0010) (Figure 6) and a corresponding strong correlation was found between the beta diversity of ‘strict’ plant pathogens and the log-transformed distance to closest mature C. scoparius (R2 = 0.53; P < 0.0001).

Proportional abundances

At the level of soil cores, the proportional abundance of Ascomycota (t = -2.871; P = 0.0041) and ‘strict’ symbiotrophs (t = -2.157; P = 0.0310) decreased with C. scoparius coverage while the proportional abundance of Mortierellomycotina (t = 2.638; P = 0.0083) and ‘strict’ plant pathogens (t = 2.767; P = 0.0057) increased with C. scoparius coverage (Table 1).

The increase in the proportional abundance of ‘strict’ plant pathogens could also be seen across plots (R2 = 0.30; P = 0.0191) and although the proportional abundance of Glomeromycotina was unresponsive to C. scoparius at the level of soil cores, the proportional abundance of Glomeromycotina was greater in plots with higher C. scoparius coverage (R2 = 0.26; P = 0.0299) (Figure 5).

For both Glomeromycotina and Mucoromycotina, there was an increase in the proportion of rarer OTUs (OTUs occurring in less than half of plots) over C. scoparius coverage per plot (Figure 7) and corresponding increases in the number of rarer OTUs over C. scoparius coverage (Appendix B10).

The effect of C. scoparius on community composition and OTU occupancy Generally, plots with high C. scoparius coverage tended to form more homogeneous communities compared with more heterogeneous communities in plots with low to mid-range C. scoparius coverage (Figure 8).

The 20 most abundant OTUs across all three plots without C. scoparius coverage were exclusively Ascomycota or Basidiomycota, however Mortierellomycotina and Mucoromycotina OTUs were within the 10 most common OTUs from all three plots with highest C. scoparius coverage (Appendix B11). Nine out of 10 of the most dominant Ascomycota in the three plots without C. scoparius remained within the 20 most dominant Ascomycota in the three plots with most C. scoparius. Among the most abundant Mortierellomycotina OTUs (Appendix B12), 14 out of 20 were shared between the three plots with highest C. scoparius coverage and three plots without C. scoparius.

The most abundant Glomeromycotina OTU belonged to the family of Glomales according to Morton and Redecker (2001). The five most abundant Glomeromycotina OTUs across the three plots without any C. scoparius coverage were all Glomeraceaea sp. from the family of Glomeraceae. In contrast, only one of the five most abundant Glomeromycotina OTUs from the three plots with highest C. scoparius coverage was a Glomeraceae sp., the other four belonging to the families of Ambisporaceae, Archaeosporaceae and Claroideoglomeraceae.

Figure 3. Gamma (plot-level, n = 18) diversity of Glomeromycotina (above) and Mucoromycotina (below) over C. scoparius % coverage. P and R2 values are given in the plots. Corresponding plots for gamma diversity over log-transformed distance from the extracted soil to the closest mature C. scoparius (mm) are given in Appendix B3.

Table 1. Linear mixed-effect model results for alpha diversity (level of soil cores) and different measurements of C. scoparius density, i.e., C. scoparius % coverage and log-transformed distance between the extracted soil core and the stem of the closest C. scoparius (mm). P value estimates are likewise given for proportional abundance and different measurements of C. scoparius density, except for all fungi (indicated by “.”). Accompanying t-values are compiled in Appendix B4. ECM = Ectomycorrhizal fungi. Arbuscular mycorrhizal fungi (AMF) is not included in the table as OTUs for AMF according to FUNGuild (Nguyen et al. 2016) matched exactly with the OTUs for Glomeromycotina according to the UNITE public database (Nilsson et al. 2018).

Alpha Diversity Proportional Abundance Broom % log(Distance Broom % log(Distance Coverage to Broom) Coverage to Broom) All fungi 0.0221 0.0025 . . Ascomycota 0.5570 0.0196 0.0041 0.4754 Basidiomycota < 0.0001 0.0013 0.1227 0.2262 Glomeromycotina 0.1350 0.9485 0.1773 0.5797 Mortierellomycotina 0.0006 0.5088 0.0083 0.5281 Chytridiomycotina 0.0001 0.0035 0.1324 0.0049 Mucoromycotina 0.4903 0.5558 0.2077 0.7447 Antagonists 0.0023 0.0573 0.3491 0.4306 Symbiotrophs 0.9183 0.0631 0.0310 0.5817 Saprotrophs 0.0464 0.0042 0.3360 0.6416 Plant pathogens 0.0026 0.0870 0.0057 0.0020 Pathogens of fungi 0.3686 0.4951 0.0908 0.7617 ECM (FUNGuild) 0.7239 0.1358 0.4264 0.9678

Figure 4. Average alpha diversity ± SE (min. 23 soil cores per plot) of all fungal OTUs over C. scoparius % coverage (above) and average alpha diversity of all fungal OTUs over the distance from the extracted soil core to the closest mature C. scoparius (below). P and R2 values are given in the plots. A plot for average alpha diversity per plot over log-transformed distance of closest C. scoparius (whether mature or immature) is given in Appendix B6.

Figure 5. [Next page] Average alpha diversity and proportional abundance of fungal OTUs (per plot) according to fungal taxa and functional traits over C. scoparius % coverage. Regression lines are shown when P < 0.05 and P and R2 values are given in the plots. All OTUs were ‘strictly’ classified into functional guilds (results with ‘loose’ classifications, i.e., with overlap, are presented in Appendix B8).

Figure 6. Beta (β) diversity of Glomeromycotina and Basidiomycota over C. scoparius % coverage (below) and Mucoromycotina and plant pathogens over C. scoparius % coverage (next page). P and R2 values are given in the plots.

Figure 6. [Continued]

Figure 7. Proportion of unique Glomeromycotina and Mucoromycotina OTUs occuring in less than half of all plots (relative to all Glomeromycotina and Mucoromycotina OTUs per plot) over C. scoparius % coverage. P and R2 values are given in the graphs.

Figure 8. Principle Coordinates Analysis (PCoA) of all soil cores (n = 431) across all plots (n = 18) coloured according to C. scoparius coverage. The community matrix for all fungal OTUs was Wisconsin-transformed prior to scaling. Rank order of the plots from lowest to highest C. scoparius cover is given at the centroid of each plot (with three plots without C. scoparius labelled “0”).

Discussion

In contrast with my hypotheses and the widespread view that biological invasions are associated with a loss of diversity, my results show that C. scoparius invasion increased fungal diversity at the average point-scale per plot (average alpha diversity), including increasing the diversity of plant pathogens.

Results differed according to at which scale the study was undertaken (Table 2) and most notably for Glomeromycotina which showed no response to C. scoparius at the level of soil cores yet increased with C. scoparius cover in all of the diversity metrics at the plot scale (average alpha diversity and beta and gamma diversity) as well as in proportional abundance and in number of rare AMF OTUs. These differences related to the scale at which the study was undertaken may indicate that relatively large increases in C. scoparius biomass are required before any noticeable effect on fungal diversity may be observed.

Response of fungal gamma and alpha diversity to C. scoparius

A possible reason why increases in both gamma and alpha diversity were observed across several fungal taxa (Table 2) could be that C. scoparius enriches the productivity of the soil environment (i.e., increases biomass generation), which can in part be supported by the observed increase in average saprotroph alpha diversity per plot. Although community productivity generally increases with the number of species in local communities (Balvanera et al. 2006, Maron et al. 2011), plots with high C. scoparius cover likely have a greater proportion of topsoil which has undergone increased C and N input, to the extent that soil beneath C. scoparius has a higher productivity than uninvaded grassland. Although I can only infer a loss in plant diversity caused by increased C. scoparius coverage, highly productive alien plant species have been known to simultaneously increase productivity even while reducing local plant species diversity (Vilà et al. 2011).

The diversity of AMF has long been a key indicator of soil productivity (Van Der Heijden et al. 1998) and AMF have an important role in increasing the competitive ability of certain invasive plants (Zhang et al. 2018). Based on the correlation of AMF diversity and invasive plant success (Zhang et al. 2018), it is probable that the elevated AMF diversity following C. scoparius invasion could be one of the factors enabling C. scoparius’ spread. This increased diversity of AMF correlated with results in Bahram et al. (2020), who likewise found that sites with higher AMF diversity harboured more saprotrophs (and plant pathogens) in the topsoil. Finding more soil saprotrophs (i.e., decomposers) surrounding C. scoparius is broadly in line with higher decomposition rates found in ecosystems with higher AMF dominance (Tedersoo and Bahram 2019). My results however differ from those in Bahram et al. (2020), who found lower fungal diversity in plant

58 monocultures, as plots in my study which had near-monocultures of C. scoparius showed increases in both gamma and average alpha diversity across several fungal taxa (Table 2).

Response of fungal beta diversity to C. scoparius

Whereas C. scoparius coverage correlated with an overall increase in alpha diversity, a general decrease in beta diversity over C. scoparius coverage was observed (Table 2). This decrease in beta diversity may suggest that C. scoparius simultaneously increased local fungal diversity but also homogenised the fungal community, resulting in more species in each soil core yet less variability across cores. Aboveground homogeneity has been linked to decreased fungal beta diversity (Zak and Willig 2004, Bachelot et al. 2016) and there has generally been a correlation between the biodiversity of groups of directly or indirectly interacting organisms (Gaston 2000, Scherber et al. 2010, Peng et al. 2019). As such, the decrease in beta diversity is probably more related to C. scoparius’ aboveground dominance (i.e., increased plant homogeneity), rather than to an increase in soil productivity induced by C. scoparius.

Response of plant pathogens to C. scoparius

Pathogens of fungi did not significantly respond to C. scoparius coverage, however increased C. scoparius coverage correlated with an increase in the alpha diversity of plant pathogens (both within individual soil cores and across plots) and a decrease in plant pathogen beta diversity (Table 2).

The diversity and composition of plant pathogens is tightly connected to plant communities (Mangelsdorff et al. 2012, Hantsch et al. 2013, Latz et al. 2016). Plant pathogen alpha diversity has been known to be influenced by plant richness, as a higher richness of host plants (i.e., a broader niche) is likely accompanied by a higher richness of plant pathogens (Bond and Chase 2002). It has therefore been proposed that a reduction in plant diversity will reduce the diversity of plant pathogens (Gossner et al. 2016), particularly in AMF dominated habitats which typically experience greater antagonism from their associated soil microbiota compared with ectomycorrhizal dominated habitats (Teste et al. 2017, Kadowaki et al. 2018). As C. scoparius formed near-monocultures in some plots, finding greater plant pathogen alpha diversity with increased C. scoparius coverage was again unexpected.

Plant pathogen alpha diversity is frequently higher in host plants with a history of agricultural use or host plants with wide geographical ranges (Mitchell et al. 2010, Kamiya et al. 2014). Plant pathogen alpha diversity has also been known to accumulate according to how long an exotic plant has been established in New Zealand (Diez et al. 2010), partly as while plants become more widespread, the plants will also have an increased probability of encountering more pathogens (Hawkes 2007). Cytisus scoparius does have a broad geographic range in New Zealand (Syrett et al. 1999), where C. scoparius was naturalized by 1872 (Owen 1998). It follows that both the geographic

59 range of C. scoparius and the shrub’s long history as an invasive plant in New Zealand could have both led to gradual accumulations in plant pathogens to the extent that the richness of C. scoparius’ pathogens supersedes the pathogen richness of uninvaded grasslands, despite the uninvaded grasslands having a higher richness of host plants (Bond and Chase 2002). This process could however work both ways to the effect that C. scoparius encounters both “friends and foes” throughout its spread. AMF associating with an invasive plant has in some cases been shown to improve plant growth and resistance (Zhang et al. 2018, Chen et al. 2019a) (although see Reinhart et al. (2017) concerning limitations). Cytisus scoparius could have accumulated different species of fungi (particularly Glomeromycotina), which may counteract antagonistic effects and thereby enable increases in multiple functional groups of fungi.

It is however unlikely that C. scoparius’ large-scale accumulation of pathogens (and potentially mutualists) over time is the sole reason as to why an increase in pathogen diversity was observed. High connectivity of invasive C. scoparius populations would also enable the spread of pathogens. A plant’s life strategy and physical size (Van der Putten et al. 1993, García‐Guzmán and Heil 2014) as well as N-fixation are factors which could have influenced the studies’ outcome. Shading by C. scoparius can be expected to create moister and more moderated conditions which would likely aid in water retention (Danner and Knapp 2003). As soil water availability is considered one of the strongest predictors of fungal richness at a global scale (Tedersoo et al. 2014) and as soil water retention increases with organic matter (Gupta and Larson 1979, Emerson 1995) (although see Rawls et al. (2003) concerning limitations), a likely increase in soil organic matter caused by C. scoparius could be accompanied by increases in fungal diversity, including plant pathogen diversity.

Response of fungal proportional abundance to C. scoparius

For two rare fungal taxa (Glomeromycotina and Mucoromycotina), there was an increase in the proportion of rarer OTUs over C. scoparius coverage, which was also reflected in an increase in gamma diversity. It was still unexpected that soil under C. scoparius should harbour more unique OTUs compared to surrounding soil with higher plant diversity, although finding a higher proportion of novel fungal OTUs in soil affected by an invasive plant (compared to uninvaded soil) has been documented (Anthony et al. 2017).

Experimental design considerations Although it was my aim to sample across a natural density gradient of an existing C. scoparius invasion rather than to study the effect of artificially created plant monocultures on soil biodiversity (e.g., Gornish et al. (2016), Gibbons et al. (2017)), a major downside of not using experimental monocultures is that I cannot be certain whether C. scoparius caused the observed changes in soil fungal communities or merely responded to existing soil fungal communities. Cytisus scoparius in

60 my field-site did however commonly occur in dense irregular patches, often near grassland uninvaded by C. scoparius. The natural distribution of C. scoparius was generally indicative of patchy seed dispersal followed by spread from dense points, which may suggest that the expansion of C. scoparius was not reliant on existing soil communities.

Microbial communities are known to differ between plant species (Innes et al. 2004), and even between genotypes within species (Kowalchuk et al. 2006). Although I sampled a typical grassland site in which C. scoparius occurs commonly across New Zealand (Bellingham and Coomes 2003), I cannot claim that C. scoparius invasion in different sites will follow the same pattern.

Metabarcoded eDNA data is known to be semi-quantitative (Martínez‐García et al. 2015) as OTU frequency does correlate to a certain extent with species relative abundance (Taberlet et al. 2018). Although the diversity and load of plant pathogens can be closely linked (Hantsch et al. 2014), plant pathogen OTU abundance should still not be strictly translated to pathogen load on plants (Torchin and Mitchell 2004). As measurements for total fungal biomass are not included within this study, I cannot interpret increases in the proportional abundance of plant pathogens and Glomeromycotina over C. scoparius coverage as increases in the total biomass of plant pathogens and Glomeromycotina, and consequently place more importance on diversity estimates.

There are many different methodological variations that might have influenced results, including the size and depth of soil cores, the decision not to remove relic DNA (Carini et al. 2020) and the number of PCR replicates undertaken for each soil extract (Dopheide et al. 2019). My next chapter on eDNA pooling will further explore how the handling of soil extracts pre-PCR impacts observed outcomes.

Of all New Zealand fungi known to associate with C. scoparius, 65.1% (according to nzfungi2.landcareresearch.co.nz) had corresponding entries in FUNGuild, implying that FUNGuild likely underestimated the number of fungal OTUs. Another limitation of existing fungal databases is that fungal species known only from sequence data are not handled well (Nilsson et al. 2019b), nor necessarily inform whether fungi (and particularly cryptic species of fungi) are native to New Zealand or not. It is therefore difficult to study whether OTUs found in the proximity of C. scoparius are introduced or native (however, see Bogar et al. (2015) concerning a possible way to assess fungal geographic origins). Although I would require an analysis of species origin for confirmation, an increase in the proportion of unique OTUs found in C. scoparius yet not in surrounding uninvaded grassland could possibly be an indicator of C. scoparius co-invading with belowground mutualists, pathogens and commensals (Nuñez and Dickie 2014).

Conclusions and applications

For plant pathogens, Basidiomycota and Chytridiomycotina, having no increase with C. scoparius coverage in gamma diversity alongside an increase in average alpha diversity per plot while beta diversity decreased with C. scoparius coverage could possibly be regarded as an indicator of coalescence between previously separated fungal communities (Rillig et al. 2015). Rather than one existing fungal community outcompeting the other (Foster and Bell 2012), a new community is formed around C. scoparius composed of fungi spreading alongside C. scoparius and the existing community in uninvaded grassland.

It is possible that invasive plants accumulate soil pathogens which inhibit native plants (Mangla et al. 2008). Knowing that C. scoparius increases the diversity of putative plant pathogens across plots was an initial surprise given that most plants in my previous soil legacy experiment (Chapter 2), including natives, benefited from being planted in soil with C. scoparius’ legacy, yet it has been observed that the diversity of putative soil pathogens is not necessarily indicative of plant growth (Hawkes 2007, Van der Putten et al. 2013).

My results highlight that studying fungal communities at different scales might lead to dissimilar outcomes and puts a spotlight on the need to have a robust experimental design prior to commencing an eDNA survey (Dickie et al. 2018, Zinger et al. 2019). My next chapter will delve deeper into the methodological processes underlying eDNA surveys.

Table 2. Summary of the effect of C. scoparius coverage on fungal diversity. Effect of C. scoparius at the scale of soil cores is measured via linear mixed-effect model estimates (with plot as a random variable). Effect of C. scoparius on the regional scale is measured across all 18 plots. ‘─’ indicates no observed correlation (i.e., P > 0.05). All fungal functional traits follow ‘strict’ classifications according to FUNGuild (Nguyen et al. 2016). When the ‘strict’ classification differs from the ‘loose’ classification, the response of the ‘loosely’ grouped functional guild is shown within parentheses. Relative richness was calculated by dividing the plot-level mean alpha diversity of a group of interest (e.g., Basidiomycota) by the plot-level mean alpha diversity of all fungi. ECM = Ectomycorrhizal fungi.

Scale of sampling plots (n = 18) Scale of soil cores (n = 431) Mean alpha Beta (β) Gamma (γ) Proportional Relative Alpha (α) Proportional (α) diversity diversity diversity abundance richness diversity abundance All fungi ↑ ─ ─ NA NA ↑ NA All fungi EXCL. ↑ ─ ↑ ─ ─ ↑ ↑ Ascomycota Ascomycota ─ ─ ─ ─ ─ ─ ↓ Basidiomycota ↑ ↓ ─ ─ ↑ ↑ ─ Glomeromycotina ↑ ↓ ↑ ↑ ↑ ─ ─ Mortierellomycotina ─ ─ ─ ─ ─ ↑ ↑ Chytridiomycotina ↑ ↓ ─ ↑ ↑ ↑ ─ Mucoromycotina ─ ↑ ↑ ─ ─ ─ ─

Antagonists (general) ↑ ↓ ─ ─ ↑ ↑ ─ (↑) Plant pathogens ↑ ↓ ─ ↑ ↑ ↑ ↑ Pathogens of fungi ─ ─ (↑) ─ ─ (↓) ─ ─ ─

Symbiotrophs ─ ─ ─ ─ ─ ─ (↑) ↓ Saprotrophs ↑ ─ ─ ─ ─ ↑ ─

ECM ─ ─ (↑) ─ ─ ─ ─ ─ (↓)

Chapter 4: Consequences of environmental DNA pooling

Abstract

DNA-based techniques are increasingly used to assess biodiversity both above- and belowground. Most effort has focussed on bioinformatics and sample collection, whereas less is known about the consequences of mixing collected environmental DNA (eDNA), post-extraction and pre-PCR. We applied varying degrees of pooling to stand-alone eDNA samples collected across a non-native plant invasion density gradient, and compared the fungal communities of pooled and unpooled samples. Pooling soil eDNA decreased observable fungal rarefied richness in our samples, led to phylum-specific shifts in proportional abundance, and increased the sensitivity of detection for the invasive plant’s overall impact on fungal diversity. We demonstrate that pooling fungal eDNA could change the outcome of similar eDNA studies where the aim is to: 1) identify the rare biosphere within a soil community, 2) estimate species richness and proportional abundance, or 3) assess the impact of an invasive plant on soil fungi. Sample pooling might be appropriate when determining larger-scale overarching responses of soil communities, as pooling increased the sensitivity of measurable effects of an invasive plant on soil fungal diversity.

Keywords

Diversity, environmental DNA, experimental design, fungal communities, metabarcoding, sampling, soil DNA extraction

Introduction

High throughput DNA sequencing technology (Caporaso et al. 2012) is increasingly used for determining the composition of ecological communities, both terrestrial and aquatic, and for testing ecological hypotheses (Holdaway et al. 2017). These approaches have the potential to revolutionize biodiversity and conservation monitoring (Lindahl et al. 2013). One technique in particular, DNA metabarcoding, can identify the presence of a multitude of species across a wide taxonomic range (Taberlet et al. 2012), which previously could only be achieved through the time- consuming morphological identification of individual organisms (Lawton et al. 1998). The growing use of DNA metabarcoding to sequence environmental DNA (eDNA, i.e., DNA extracted from soil, water, air or other substances) has brought to attention the need for in-field collection and sampling protocols as well as instructions on how to process obtained samples in the laboratory environment. There have been previous reviews of metabarcoding methods which focus on statistical replication in sampling (Lennon 2011), the processing of collected samples (Lear et al. 2018), as well as data reporting and bioinformatics analysis (Hiraoka et al. 2016). However, there are very few DNA metabarcoding studies on the effect of mixing extracted eDNA samples together, i.e., “pooling” samples prior to being sequenced. The most thorough study to date on determining the effect of eDNA pooling is possibly Avis et al. (2010), who mixed up to 20 pre- selected fungal species prior to molecular analysis. Given how common pooling is when undertaking community studies (Dickie et al. 2018) and given that sampling and subsequent pooling techniques are the basis on which valid inferences are dependent (Crawley 2015), the consequences of pooling large numbers of eDNA samples prior to sequencing deserves more attention.

Sample pooling is typically performed as a means of estimating the dominant species in a given area (Ellingsøe and Johnsen 2002), or in order to gather more information on the complexity of sampled biodiversity. In either case, pooling is closely associated with a loss of spatial variability information (Dickie et al. 2018), which can be of minor or major concern dependent on the research question. Two seemingly unavoidable downsides of pooling are a reduced ability to detect rare species, particularly for fungi in comparison to bacteria (Manter et al. 2010), as well as a reduced ability to estimate species richness (Kang and Mills 2006). A methodological issue surrounding the detection of rare species occurs at the PCR amplification stage of eDNA studies. As PCR is a competitive process (Siebert and Larrick 1992) where the level of amplification achieved is positively correlated with the amount of starting template DNA, species with relatively low abundance will undergo PCR amplification to a lesser degree than species with a relatively high

65 abundance. Pooling is likely to dilute rare DNA templates to the extent that amplification of rare templates may be insufficient for detection.

There are studies which suggest that pooling pre-PCR has little effect on the perceived community (Manter et al. 2010, Osborne et al. 2011) making sample pooling economically advantageous, yet it has also been observed that pooling decreases detected variability when compared to not pooling (Osborne et al. 2011). When presented with a pooled sample, it is possible to lessen the issue of decreased detected variability by amplifying several diluted subsets of the pooled sample (which would increase the likelihood that less abundant species are successfully amplified and detected), yet in such cases it might have been better to have instead taken multiple stand-alone samples allowing for additional spatial variability analyses (Dickie et al. 2018). This is because a rare species in a given area can be 1) found in low abundance but ubiquitously distributed, or 2) found in high abundance at fine scales but heterogeneously distributed (Green et al. 2004). Such spatial distributions could be characteristic to different fungal taxa, for instance, it has been observed that wood-inhabiting members of the fungal phylum Ascomycota are more specific to certain tree species compared with the more homogeneously distributed fungal phylum Basidiomycota (Purahong et al. 2018).

Among studies which have directly explored the consequences of pooling eDNA samples (Ellingsøe and Johnsen 2002, Manter et al. 2010, Osborne et al. 2011, Song et al. 2015, Sato et al. 2017), only Manter et al. (2010) and Song et al. (2015) deal with soil-extracted fungal samples. Song et al. (2015) used two pooled samples in their experiment, each created by mixing four soil samples before eDNA extraction; they observed that although pooled samples had a higher fungal OTU richness compared with stand-alone samples, computational pools created by combining stand- alone samples showed higher OTU richness compared with the physical pool. In the case of Manter et al. (2010), a community fingerprinting (ribosomal intergenic spacer analysis) approach was adopted using soil from three very dissimilar sampling sites from both hemispheres; they found that in their sample sets, fungi were typified by locally abundant but spatially rare phylotypes, whereas bacteria in their study were typified by locally rare but spatially ubiquitous phylotypes. As a result, pooling would differentially influence their plot comparisons and would mask a significant proportion of their detectable microbial community, particularly for fungi due to their higher spatial heterogeneity. Sato et al. (2017) tested whether pooling of eDNA samples from four Japanese lakes could be used to evaluate the biodiversity of freshwater fishes and found that their pooling strategy was unsuitable for estimating species richness, yet had potential for among-site comparison of their fish communities. Osborne et al. (2011) studied how their pooling strategies influenced the detected composition of bacterial communities in three distinct Australian land-use plots, finding that as few as 8 or 10 cores per plot was sufficient to detect significant differences between the bacterial communities from their three study sites.

In a study on soil bacterial communities dug up from a Danish forest and identified via a denaturing gradient gel electrophoresis approach (DGGE), Ellingsøe and Johnsen (2002) observed that using larger soil samples (which to a certain degree are comparable with pooled samples) could be more appropriate when studying anthropogenic activities on bacterial community structure when compared to smaller soil samples where chance variations could play a larger role. Soil resources (e.g., organic matter) and soil properties can often vary at a smaller scale than sample volume (Cappai et al. 2017, Evgrafova et al. 2018) and could be considered a source of unexplained variance which may conceal the larger-scale effect of environmental drivers (e.g., anthropogenic effects, invasive species impacts). In relation to fungi, the relative abundance of the fungal phylum Ascomycota has been shown to be negatively correlated with soil organic C in agricultural fields, while the relative abundance of the fungal phyla Basidiomycota is positively correlated with soil organic C (Zebarth et al. 2018). In this example, pooling samples could have a ‘double-edged sword’ effect in terms of studying the soil community: on the downside, sample pooling could dampen the smaller-scale effect of soil organic C and distort measurements of relative abundance, yet on the upside enhance the detection of an overarching effect such as anthropogenic activity.

Although some valuable insights on soil eDNA pooling have been provided via a small number of mixed samples pooled pre-extraction (Song et al. 2015), to the best of my knowledge, there is yet no comprehensive study which examines how sample pooling post-extraction affects the species richness and proportional abundance measurements of fungal eDNA, nor how identifying the presence of an ecological gradient (e.g., the effect on an encroaching invasive plant species on belowground species richness) might be hindered or exaggerated by pooling eDNA samples. Given the cost of field sampling, wet-lab processing and sequencing, it is desirable to neither under- nor oversample when conducting an eDNA-based ecological survey. Methodologically sound eDNA sample preparation is the foundation for subsequent analyses and examining the consequences of sample pooling would be one way to help assess how reliable and reproducible an eDNA survey is. In this study, I applied four degrees of eDNA pooling to individual soil core extracts collected from six plots along an exotic plant’s invasion gradient, followed by Illumina sequencing of indexed PCRs targeting the ribosomal internal transcribed spacer (ITS) region (Schoch et al. 2012) within soil fungi.

My aim was to investigate how varying degrees of eDNA sample pooling affects species richness and proportional abundance of soil organisms, specifically fungi, and to examine whether sample pooling enhances or dampens the overarching effect of an exotic plant (C. scoparius) on soil fungal communities. Considering the large variability of data curation steps found in eDNA analyses (Dickie et al. 2018, Calderón‐Sanou et al. 2019), species richness estimations have been known to be sensitive to inaccuracies (Flynn et al. 2015, Dopheide et al. 2019), yet in contrast, studies which focus on comparing communities’ composition are considered less prone to errors (Leray and

Knowlton 2015, Taberlet et al. 2018). I therefore chose to focus on how pooling affects proportional abundances within my sampled communities, particularly at the level of fungal phylum, as this observation would likely be more relatable to other fungal eDNA studies compared with species richness.

As a broader area would have been covered, it is reasonable to expect a pooled eDNA sample to have a higher species richness than any stand-alone sample that the pooled sample is composed of, however, I surmised this to be at the cost of a reduced detectability of the rare biosphere.

I hypothesised that:

• Although pooling eDNA samples is expected to increase species richness, a proportion of the “true” species richness is likely to become undetectable by DNA metabarcoding techniques because of pooling, partly due to over-dilution of rare DNA templates.

• Given that different fungal phyla are known to have broader or more restricted distributions (Purahong et al. 2018, Zebarth et al. 2018), sample pooling will cause distortions in the “true” proportional abundance of fungal taxa.

• As sample pooling will likely decrease within-plot variation of fungal communities, caused by multiple unaccounted abiotic and biotic factors, the use of pooled samples should therefore be more sensitive to larger scale between-plot comparisons, which in this case is the presence of an exotic plant (Cytisus scoparius) across an invasion gradient.

To test these hypotheses, I apply varying degrees of pooling to stand-alone eDNA samples systematically collected across an exotic plant’s invasion gradient and compare the fungal communities of computationally pooled samples with my physically pooled samples (pooled after DNA extraction).

Methods

Study site and field experiment

The study site was located in the Saint James conservation area in New Zealand’s South Island (- 42.460273 Lat., 172.830938 Long.; elevation = 800–900 m.a.s.l.; mean annual temperature = 10.3°C; mean annual rainfall = 1158 mm, Hanmer forest weather station). Cytisus scoparius (Scotch broom) is widely spread throughout this region and a description of the site’s vegetation is given in Broadbent et al. (2017). Permanent 20 × 20 m vegetation plots were laid out at the site by Manaaki Whenua – Landcare Research, following standard field protocols (Hurst and Allen 1993). For this experiment, I selected six permanent vegetation plots across a C. scoparius density gradient (i.e., two plots with low C. scoparius coverage, two plots with intermediate C. scoparius coverage and two plots with high C. scoparius coverage) and all plots were located within 1 km of each other. Field sampling took place from 14 February 2017 to 17 April 2017. For each of the six permanent vegetation plots, 24 individual georeferenced soil cores were taken, totalling in 144 spatially explicit soil samples. The six plots were a subset of those used in my natural survey of C. scoparius invasion (Chapter 3) and subsequent methods regarding how I undertook in-field soil sampling as well as measured C. scoparius coverage are identical to the methods in Chapter 3.

Figure 1 gives an overview of how each soil sample was processed, once brought back from the field site. Each ~250 g soil sample (stored at 4°C) was broken up manually and spread out evenly on clean paper. Using bleached forceps and spatulas, a ~10 g mixed soil sample was obtained by systematically extracting 10 × ~1 g subsamples from across the initial sample. The mixed subsampled soil did not contain any roots more than 5 mm in width or stones larger than 5 mm in diameter and obvious insects (e.g., ants, larvae) were avoided. The processed soils were kept frozen at -18°C until DNA extraction.

Experimental design and wet-lab processing

Both the kit used for soil DNA extraction and the chosen fungal primers were recommended by Lear et al. (2018). DNA extraction was performed on the 144 soil cores using DNeasy PowerSoil® HTP 96 Kits (Quiagen), according to the manufacturer’s instructions and loading the maximum amount of recommended soil for DNA extraction (250 mg). As part of the PowerSoil® protocol, mechanical lysis of the soil samples was performed using a Spex® Sample Prep 1600 MiniG. Five μL subsamples of the 144 stand-alone soil extracts (at full concentration) were mixed in equal proportions to create pooled soil samples, composed of 3, 6, 12 or 24 combined soil extracts. I chose the pooling partitions in such a way that smaller partitions “fitted” into larger ones, creating “pools within pools” to enable more practical comparisons. Figure 2 shows the partitions based

69 on which the individual soil extracts were pooled together. To avoid potential pipetting errors, pools of three samples were first created (8 × 3-sample pools per plot), which were then used to produce pools of six samples (4 × 6-sample pools per plot) by combining equal quantities of two 3-sample pools. Pools of six were in turn mixed together to create pools of 12, which were finally combined to produce a pool of all 24 samples extracted from a plot. Following this method, a sum of 15 pooled samples were created per plot (alongside the 24 individual samples per plot) and 90 pooled samples were prepared in all along with 144 individual samples. A total of 234 eDNA extracts (both pooled and individual) then underwent the same following PCR amplification steps.

Based on amplification protocols outlined by the Earth Microbiome Project (Gilbert et al. 2014) (http://press.igsb.anl.gov/earthmicrobiome/protocols-and-standards/its/), two single-indexed DNA libraries were assembled from the 188 individual soil extracts and 90 pooled soil extracts using the fITS7 general fungal primer (5′- GTG ART CAT CGA ATC TTT G -3′) (Ihrmark et al. 2012) and the ITS4 reverse primer (5’- TCC TCC GCT TAT TGA TAT GC -3’) (White et al. 1990). The ITS4 reverse primer was designed with both Illumina adapter sequences and index sequences (Caporaso et al. 2011), permitting future identification of the sequenced amplicons. The fITS7 primer (along with its Illumina adapter sequence) was ordered from “Integrated DNA Technologies” (Purification method: Standard Desalting). The Illumina adapter for fITS7 was 5’- AAT GAT ACG GCG ACC ACC GAG ATC TAC AC -3′ and the Illumina adapter for ITS4 was 5’- CAA GCA GAA GAC GGC ATA CGA GAT -3′.

The amplification experiments were performed using an Eppendorf vapoprotect Mastercycler® in a 25 µL mixture volume containing 0.2 μL FastStart™ DNA polymerase (Merck), 0.5 μL dNTP mixture (10 mM each), 2.5 μL PCR buffer (with 20 mM MgCl2, sourced from Merck), 2 μL 2.5 μM of each forward and reverse primer, 1.25 μL 10 μM molecular grade Bovine Serum Albumin, 1 μL 10× diluted DNA template (same dilution for both pooled and individual soil eDNA extracts) and 15.55 μL filtered deionized water (obtained via Milli-Q® water purification system and filtered through a Biopak® Polisher). Bovine Serum Albumin was used to reduce the effect of PCR inhibitors derived from soil (Jiang et al. 2005). All PCR reagents prior to adding the DNA template were assembled in a dedicated UV light irradiated chamber with a dedicated set of micropipettes. PCRs were carried out under the following conditions: a denaturation step of 5 min at 94°C, followed by 35 cycles of 30 s at 94°C, 30 s at 57°C and 30 s at 72°C, with a final step at 72°C for 7 min (and held at 4°C). All PCRs were carried out in duplicate along with positive and negative controls. Agarose gel electrophoresis, stained with RedSafeTM (iNtRON) and using a 1% agarose gel, was performed on the PCR product to confirm amplification. No DNA was observed in negative controls and samples which showed poor amplification were rerun. The 10× dilution of the DNA templates (with filtered deionized water) was undertaken after initially performing PCRs

70 with 1 μL of the undiluted PowerSoil® DNA elution (both for pooled and individual samples), which yielded less PCR product according to gel runs, possibly due to PCR inhibition caused by soil components such as tannins (Kreader 1996). As the PCRs were performed in duplicate, 20 μL of each duplicate PCR product were combined prior to normalization.

Following manufacturer’s instructions, SequalPrepTM Normalization Plates (ThermoFisher Scientific) were used to both clean the obtained amplicons and to normalize their concentration relative to each other. Twenty-five microlitres of mixed duplicate PCR product (25 μL being the recommended maximum) underwent normalization as part of the SequalPrepTM protocol. I eluted my normalized PCR product with 12 μL elution buffer (instead of 20 μL recommended by the manufacturer) due to a previous unsuccessful sequencing library submission that might possibly have been caused by too low a concentration of PCR amplicons. The resulting concentration of normalized samples was 2.14 ng/μL. Two sequencing libraries were prepared; for each library, 12 μL of normalized indexed PCR amplicons were pooled together, vortexed and 110 μL each of the two resulting mixtures were sent to Massey Genome Service, New Zealand, to undergo Illumina MiSeqTM (run option: 2×250 base PE v2) (Caporaso et al. 2012).

Bioinformatics and statistical analysis

I used R version 3.5.0 (Team 2013) for creating graphs and conducting analyses. The subsample size for rarefying my community was set to the minimum number of sequences of any sample. I

71 created in silico samples which were pooled computationally to correspond with my physically pooled samples via the aggregate function in “stats (v3.62)” (Team 2013). When examining the composition of my samples according to fungal phylum, I generated randomly rarefied community versions of my dataset via the rrarefy function in “vegan” (Oksanen et al. 2013) before measuring the proportion belonging to a specific fungal phylum. This process was iterated 100 times and a mean proportion was calculated for each fungal phylum. The rrarefy function was here appropriate as random rarefaction is performed without replacement so that the variance of community metrics is not related to the size of the sample.

To quantify the effects of pooling on fungal rarefied richness and proportional abundance, I used linear mixed-effect models via the R package “lme4 (v1.121)” (Bates et al. 2014), setting sampling plot as a random effect. The mixed models break down the variance into inter- and intraplot components and thus improve the true structure of the randomness present in the data (Millar and Anderson 2004). When fitting my linear mixed-effect models to rarefied richness over C. scoparius % coverage, I optimised my model using the lme4’s update function, reducing the model by excluding any interaction when the P value for the interaction term was higher than 0.05. I compared my simplified models to my original models using lme4’s anova function to make sure that the interaction between both models was non-significant (P > 0.05).

Figure 1. Overview of sample collection from each field plot and laboratory processes used in the study.

Figure 2. Partitions designed for the pooling of soil eDNA extracts. Individual cores are shown in green, pooled eDNA extracts comprised of the individual cores are shown in purple. Smaller pooling partitions were designed to fit into larger partitions.

Results

In total there were 3471 fungal OTUs, the three most dominant fungal phyla and subphyla being Ascomycota (2039 OTUs, 58.7%), Basidiomycota (1048 OTUs, 30.2%) and Mortierellomycotina (119 OTUs, 3.4%) (Appendix C1). Appendix C2 shows a summary of the mean number of OTUs according to fungal phyla and pooling treatment.

Effect of pooling on richness

Across all six plots, physcially pooling eDNA samples prior to PCR amplification increased the rarefied richness of pooled samples compared to individual samples, yet the rarefied richness was highest for each plot when the individual samples were computationally combined to correspond with the pooled samples (Figure 3). The higher the degree of physical pooling (i.e., the more stand- alone samples a pooled sample was initially composed of), the higher the loss in rarefied species richness when compared to individual samples pooled computationally. Compared with Basidiomycota and Mortierellomycotina, the loss in rarefied richness was most pronounced within the fungal phylum Ascomycota (Figure 4).

Effects of pooling on composition

Again across all six plots, pooling eDNA samples shifted the proportional abundance of three out of six tested fungal taxa, which consisted of the most dominant fungal phyla and subphyla in the dataset (Figure 5). Any degree of eDNA pooling resulted in a downwards shift in the proportional abundance of Ascomcota and augmented that of Basidiomycota and Mortierellomycotina. Pooling showed no apparent effect on the proportional abundance of Glomeromycotina, Chytridiomycota or Mucoromycotina, although a weak increasing trend in the proportional abundance of Mucoromycotina (t = 0.784, P = 0.0716) occurred when pooled. The calculations for proportional abundance were redone without rarefaction (flowchart in Appendix C7) and no qualitative difference in results was found.

Fungal phyla which had a lower rank abundance in stand-alone eDNA samples (i.e., occurred in- field at relatively lower frequencies and/or were sequenced to a lesser degree), were less detectable the higher the degree of pooling (Table 1); yet the proportion of Mortierellomycotina OTUs increased the higher the degree of pooling (Appendix C8), which could also be observed for each of the 6 plots individually (Appendix C8). A similar general pattern was observed when looking at propotional rank sequence abundance both of all plots simultaneously (Appendix C9) and of each plot individually (Appendix C10). Whereas stand-alone samples were dominated by Ascomycota, when physically pooled, Basidiomycota and Mortierellomycotina both increased in rank sequence abundance. When considering the OTUs with the highest rank abundance within all individual

75 samples (n = 143), 9/10 were Ascomycota, yet only 2/10 of the most abundant OTUs were Ascomycota when analysing all pooled eDNA samples (n = 90).

Effects of pooling on perceived impact of C. scoparius

Cytisus scoparius coverage increased rarefied diversity for all fungi and for the three most abundant fungal phyla independently. Different levels of pooling showed dissimilar responses of rarefied fungal diversity to C. scoparius coverage (Figure 6), with higher levels of pooling showing a steeper upward trend for all fungi, for Basidiomycota and Mortierellomycotina, yet not for Ascomycota. When considering individual samples, C. scoparius coverage had little effect on the rarefied richness of Mortierellomycotina, yet when considering pooled samples, the rarefied richness of Mortierellomycotina increased with C. scoparius coverage (Figure 6).

As C. scoparius coverage increased, the rarefied richness of Basidiomycota, the 2nd most abundant fungal phylum, decreased in proportion to all fungal phyla when pooled computationally yet increased when pooled physcially (Figure 7). Also, as C. scoparius coverage increases, the proportion of Mortierellomycotina relative to all fungal phyla decreased except for when the samples were pooled by 12 or pooled by 24, in which case C. scoparius coverage has no considerable effect (Figure 7).

Pooled × Pooled log(samples in pool) log(samples in pool) Rarefied richness of all fungal phyla < 0.0001 < 0.0001 < 0.0001 Rarefied richness of Ascomycota < 0.0001 < 0.0001 < 0.0001 Rarefied richness of Basidiomycota < 0.0001 < 0.0001 < 0.0001 Rarefied richness of Mortierellomycotina < 0.0001 < 0.0001 < 0.0001

Figure 3. Rarefied richness over number of samples per pool (note log scale axis) for all fungal phyla. Purple triangles denote the physical pooling of samples whereas green circles represent the corresponding individual samples, the rarefied diversity of which has been pooled computationally to correspond with the physically pooled samples. Individual plot numbers are indicated on the top left of each graph and lines follow linear mixed-effect model fit. The p-value estimates in the linear mixed-effect model are presented in the below table (accompanying t-values are compiled in Appendix C3).

Figure 4. [Next page] Rarefied richness over number of samples per pool for individual fungal phylum. Purple triangles denote the physical pooling of samples whereas green circles represent the corresponding individual samples, the rarefied diversity of which has been pooled computationally to correspond with the physically pooled samples. Individual plot numbers are indicated on the top left of each graph and lines follow linear mixed-effect model fit. The p-value estimates in the linear mixed-effect model are presented in the above table (accompanying t-values are compiled in Appendix C3).

Figure 5. [Next page] Proportional abundance by fungal phylum over number of samples per pool (note log scale axis). Purple triangles denote the physical pooling of samples whereas green circles represent the corresponding individual samples, the proportional abundance of which has been pooled computationally to correspond with the physically pooled samples. Individual plot numbers are indicated on the top left of each graph and lines follow linear mixed-effect model fit. The p-value estimates in the linear mixed-effect model are presented in the above table (accompanying t-values are compiled in Appendix C4).

Pooled × Pooled log(samples in pool) log(samples in pool) Proportional abundance of Ascomycota < 0.0001 0.7390 0.6559 Proportional abundance of Basidiomycota < 0.0001 0.8656 0.8175 Proportional abundance of Mortierellomycotina < 0.0001 0.8593 0.8123

Dominant Core Rare Total Ascomycota 323 296 277 896 Basidiomycota 124 134 146 404 Chytridiomycotina 0 4 0 4 Glomeromycotina 4 7 10 21 Mortierellomycotina 21 22 20 63 Stand-alone Mucoromycotina 8 12 21 41 Unidentified 0 8 5 13 Ascomycota 171 159 154 484

Basidiomycota 59 66 68 193 3 Chytridiomycotina 0 0 0 0 Glomeromycotina 0 4 6 10 Mortierellomycotina 12 9 14 35 Pooled by Pooled Mucoromycotina 3 5 0 8 Unidentified 0 0 3 3 Ascomycota 141 129 133 403

Basidiomycota 49 55 54 158 6 Chytridiomycotina 0 0 0 0 Glomeromycotina 0 4 5 9 Mortierellomycotina 11 9 7 27 Pooled by Pooled Mucoromycotina 0 4 3 7 Unidentified 0 0 0 0 Ascomycota 125 107 119 351 Basidiomycota 38 49 46 133 12 Chytridiomycotina 0 0 0 0 Glomeromycotina 0 0 3 3 Mortierellomycotina 9 10 3 22 Pooled by Pooled Mucoromycotina 0 4 3 7 Unidentified 0 0 0 0

Ascomycota 89 85 79 253 Basidiomycota 28 34 32 94 24 Chytridiomycotina 0 0 0 0 Glomeromycotina 0 0 0 0 Mortierellomycotina 7 5 8 20 Pooled by Pooled Mucoromycotina 0 0 3 3 Unidentified 0 0 0 0

Table 1. Taxonomic composition of OTUs across all plots, split into the top 3rd (“dominant”), middle 3rd (“core”) and bottom 3rd (“rare”) proportional rank abundance percentile for each degree of pooling. Note that “0” values are in bold to accentuate the loss of rare fungal taxa the higher the degree of pooling. The same data is visualised in Appendix C8 in terms of proportions.

Broom Broom coverage × Samples in pool × coverage Samples in pool Pooled samples in pool pooled Rarefied richness of all fungal phyla 0.02558 < 0.0001 < 0.0001 0.02870 < 0.0001

Rarefied richness of Ascomycota 0.06400 < 0.0001 < 0.0001 • < 0.0001 Rarefied richness of Basidiomycota 0.01834 < 0.0001 < 0.0001 0.00981 < 0.0001 Rarefied richness of Mortierellomycotina 0.44781 < 0.0001 < 0.0001 0.01559 < 0.0001

Figure 6. Rarefied richness over mean C. scoparius % coverage for all fungal phyla (above) and for individual fungal phylum (next page). Lines follow linear mixed-effect model fit. The p-value estimates in the linear mixed-effect model are presented in the above table (accompanying t-values are compiled in Appendix C5).

Figure 6. [Continued]

Rarefied richness of Broom Samples Broom coverage × Broom coverage × Samples in pool × Broom coverage × coverage in pool Pooled samples in pool pooled pooled samples in pool × pooled Basidiomycota 0.294 < 0.0001 < 0.0001 0.205 0.00382 < 0.0001 0.0372 ~ of all fungal phyla Mortierellomycotina 0.04234 0.80651 0.01272 0.09242 . < 0.0001 . ~ all fungal phyla Ascomycota 0.9277 0.1055 < 0.0001 . . . . ~ all fungal phyla

Figure 7. Rarefied richness of individual fungal phyla in proportion to the rarefied richness of all fungal phyla over mean C. scoparius % coverage. Lines follow linear mixed-effect model fit. The p- value estimates in the linear mixed-effect model are presented in the above table (accompanying t- values are compiled in Appendix C6).

Figure 7. [Continued]

Discussion

The results of this study suggest that, compared with computationally pooled soil eDNA extracts, physically pooling soil eDNA pre-PCR will 1) decrease observable fungal rarefied richness, 2) lead to fungal phylum-specific shifts in proportional abundance and 3) increase the sensitivity regarding how an invasive plant’s overarching impact on fungal diversity is detected. Pooling fungal eDNA might change the outcome of similar eDNA studies where the aim is 1) to identify the rare biosphere within a soil community, 2) to estimate species richness and proportional abundance, or 3) to assess the impact of an invasive plant on soil fungi. In more detail, compared with computationally pooled soil eDNA extracts, physically pooling soil eDNA pre-PCR will 1) decrease observable fungal rarefied richness, particularly for Ascomycota in relation to Basidiomycota and Mortierellomycotina, 2) lead to fungal taxa-specific shifts in proportional abundance, which increases for Basidiomycota and Mortierellomycotina at the expense of Ascomycota, and 3) increase the sensitivity as to what extent an encroaching invasive plant (C. scoparius) increases the rarefied richness of all fungi as well as the richness of individual fungal taxa. Moreover, the effect of C. scoparius on the relative rarefied richness of Basidiomycota (in proportion to all fungi) can be influenced by whether or not an eDNA study uses sample pooling.

Effect of pooling on richness

Soil commonly harbours a large diversity of microorganisms in close proximity as a function of physical characteristics (Kang and Mills 2006). As shown in my results on rarefied richness over the number of samples per pool, it was therefore expected that combining multiple samples captures more OTU richness compared to individual samples (Song et al. 2015). Even though this increase in rarefied richness was expected, physically pooled fungal eDNA has been known to have a lower fungal species richness when compared to individual samples (Branco et al. 2013), despite a pooled sample encompassing a larger area. This decrease in species richness is due to rare taxa being not well-represented in pooled samples (Ohman and Lavaniegos 2002) and my results show that minor levels of physical pooling (3 stand-alone samples per pool) risk reducing the rarefied richness of Ascomycota in particular.

Effects of pooling on composition

Given the shifts in proportional abundance observed in my data, where pooling decreased the proportional abundance of the community’s most dominant fungal phylum (Ascomycota) in favour of less common taxa (Basidiomycota and Mortierellomycotina), it may be proposed that individual OTUs of Ascomycota are spatially rare but locally dominant. On the other hand, OTUs of Basidiomycota and Mortierellomycotina occur more homogenously distributed in a given area,

87 yet with lower dominance (a general overview of the occupancy of individual fungal OTUs by phyla is given in Appendix C11). Although restricted to a broad observation, such differing fungal spatial distributions could be responsible for why pooling stand-alone eDNA rich in locally dominant Basidiomycota would dilute the more homogeneous Ascomycota, thus making them less detectable by metabarcoding techniques as well as increasing the proportional abundance of Basidiomycota while decreasing that of Ascomycota.

Effects of pooling on perceived impact of C. scoparius

The significant interaction of C. scoparius cover and perceived species richness suggests that pooling is a useful technique when studying overarching large-scale effects such as the impacts of an invasive plant on soil communities, as suggested by Ellingsøe & Johnsen (2002). Pooling reduced the variability between samples, thereby providing a more general expression of the overall community structure in a given plot (Ellingsøe and Johnsen 2002, Manter et al. 2010), yet at the expense of within-plot accuracy obtained by using a higher number of samples (Ranjard et al. 2003). When considering how using pooled samples showed an increase in the proportion of Basidiomycota (relative to all fungi) as C. scoparius increases whereas stand-alone samples showed a decrease, the question remains as to which is the ‘real’ process taking place, yet it is evident that dissimilar (and in this case mutually exclusive) broad ecological effects can be observed according to sample processing.

Experimental design considerations

There are multiple biases surrounding PCR (Acinas et al. 2005) and regulating the amount of DNA template for PCR has been known to improve PCR efficiency (Wilson 1997, Lindahl et al. 2013). Although all PCRs were performed in duplicate to reduce bias in template-to-product ratios (Polz and Cavanaugh 1998), the methods do not take into account how the concentration of the soil eDNA pre-PCR might have impacted the composition of the physical pools (e.g., some soil extracts could have had a much lower soil eDNA concentration from the offset). I did not quantify the soil eDNA extract as it would have been very difficult to determine how much of the extracted DNA was derived from fungi, yet observing similar patterns across six plots should account for any differences in the relative concentration of my pooled samples pre-PCR.

I pooled DNA extracts rather than soils. Instead of extracting DNA from 0.25g of soil per collected sample, if I had instead sampled 0.25g of soil from a pool of multiple (i.e., up to 24) soil cores, this would almost certainly have caused a very large variability between samples which I wanted to avoid.

Conclusions and applications

Pooling is a common practice in eDNA studies (Dickie et al. 2018), both at the level of plot (Osborne et al. 2011) or within subset categories such as soil depth (Tveit et al. 2013). Although there have been two notable cases where pooling before or after PCR was observed to have little effect on the perceived community (Manter et al. 2010, Osborne et al. 2011), in both cases this was specific to bacteria as opposed to fungi and it has been suggested that fungi can be more susceptible to pooling due to having a more spatially heterogeneous distribution compared to bacteria (Manter et al. 2010). Such a patchy fungal spatial heterogeneity, which has long been observed at fine scales (Horton and Bruns 2001), could be an underlying factor causing locally dominant but spatially rare taxa to become overly diluted in pooled eDNA samples, rendering them untraceable to metabarcoding techniques.

The decision whether or not to pool can be highly context-dependent and relies on weighing up costs of field sampling, DNA extraction, wet-lab processing and the sequencing technology (or other approach) used to identify the samples. The spatial heterogeneity of the studied organisms as well as the trade-offs between increased replication and improved precision per replicate need to be likewise taken into account (Dickie et al. 2018). It is uncommon in eDNA studies that multiple stand-alone samples are taken within a statistical replicate (Dickie et al. 2018), even so, such an approach should be encouraged if both within and between plot variability in community composition is to be examined (Drummond et al. 2015, Navarrete et al. 2015).

When considering the possible benefits of pooling, particularly in relation to how C. scoparius coverage in my dataset increases fungal rarefied richness, it is important to note that the intra-plot variance which pooling decreases is not necessarily ‘distracting noise’ (Ranjard et al. 2003), but potentially valuable ecological information. However, if the goal of a study is to test large scale patterns, unexplained intra-plot variance can be an obvious hurdle. With this in mind, I have two recommendations regarding the use of eDNA pooling:

• If the objective of an eDNA survey is study the rare biosphere or the proportional abundance of fungi in a given environment, then my results support Lear et al. (2018)’s recommendation that sample pooling should be avoided in favour of analysing more small subsamples and it can be added that this is particularly applicable if the fungal phyla of interest is Ascomycota, Mucoromycota, Glomeromycotina or Chytridiomycota.

• If however the objective of an eDNA survey is to study a large scale overarching effect across several plots such as that of an invasive plant species on fungal communities, then my results support Ellingsøe and Johnsen (2002)’s recommendation to use larger sample sizes, or in my case pooled samples, as these reduce intra-plot variation allowing broader-

scale effects to override the abundant fine-scale variation present in fungal communities (Dickie et al. 2002, Tedersoo et al. 2003, Taylor et al. 2014).

I would add that pooling is most amenable to Mortierellomycotina and Basidiomycota and can still be a cost-effective way to calculate species richness across larger areas, but is not as effective as multiple stand-alone samples. A future research avenue would be to examine the potential of eDNA pooling as a means of cost-effectively detecting changes in soil fungal composition caused by large-scale events such as anthropogenic activity or the spread of an invasive species.

Chapter 5: The fungal component of the soil legacy of Cytisus scoparius

Abstract

Soil legacy effects can be an important contributing cause of plant rarity and invasiveness in communities. However, the identity and composition of soil biota which may underlie plant growth responses and nutrient acquisition are rarely accounted for, even though soil microbiota can determine the availability of many essential plant nutrients or adversely stunt plant growth through antagonism, among other effects. I aimed to examine whether the soil legacy of Cytisus scoparius (Fabaceae) could be explained in terms of fungal communities. I used soil with known fungal community composition extracted from across a density gradient of exotic C. scoparius to test whether the fungal component of the soil legacy of C. scoparius favoured the growth and nutrient acquisition of a selection of plants native and exotic to New Zealand which were either able or unable to fix nitrogen. I found that of all tested fungal predictors, increased arbuscular mycorrhizal fungi richness induced by the soil legacy of C. scoparius favoured the growth of Fabaceae, particularly exotic Fabaceae, compared with non-N-fixing native plants. My results suggested that exotic Fabaceae may form interaction with a higher richness of arbuscular mycorrhizal fungi than native plants, which may underlie their invasiveness.

Keywords

Arbuscular mycorrhizal fungi, Cytisus scoparius, environmental DNA, functional traits, fungal communities, metabarcoding, mutualisms, soil legacy

Introduction

Plant-soil feedback, the process whereby a plant’s effect on soil biotic and abiotic properties influences the growth of future generations of plants, is generally considered to be an important contributing cause of plant rarity and invasiveness in communities (Klironomos 2002). In Chapter 2, I performed a greenhouse experiment examining the effects of soil under various levels of an invasive leguminous shrub (Cytisus scoparius). For that greenhouse experiment, I dug up soil from 18 permanent vegetation plots across a natural C. scoparius density gradient. The vegetation of the 18 plots ranged from grassland uninvaded by C. scoparius to near-monocultures of C. scoparius. I then cultivated 16 plant species in the 18 collected soils for ~7 months in a greenhouse environment (total number of plants = 288). At harvest, I measured shoot height, root and shoot dry biomass and obtained measurements for shoot % N and shoot % P. After plant morphometric and nutrient responses were analysed with accompanying data on soil chemistry from each of the 18 sampling plots (e.g., Olsen P, Ca, Mg), the results suggested that although some soil chemical traits had a slight correlation with the effect of C. scoparius coverage on plant growth, a portion of the effect of C. scoparius coverage remained unexplained (Table 4; Chapter 2).

Soil microorganisms often have significant positive and negative effects on plants through root- rhizosphere mutualism (Brundrett 1991), pathogen effects (Burdon 1993, Packer and Clay 2000) and by driving nutrient cycles (Crowley et al. 1991). In Chapter 3, I set out to measure part of the biological effect of C. scoparius by undertaking a natural survey on how fungal communities surrounding C. scoparius differed according to varying degrees of C. scoparius invasion. I sampled soil from the same plots which I used for my greenhouse experiment (Chapter 2) and determined that C. scoparius caused an unexpected increase in the diversity of several fungal taxa and functional groups (summarized in Table 2; Chapter 3).

Soil legacy studies are typically performed by studying plant growth in soil moulded by a previous plant (e.g., Grove et al. (2015)) or soil which has undergone a particular treatment (e.g., Pernilla Brinkman et al. (2010)). Although measuring soil chemical attributes is relatively common in soil legacy studies, the identity and composition of soil biota which may underlie plant growth responses (or nutrient acquisition) are rarely accounted for, even though soil microbiota can determine the availability of many essential plant nutrients (Edwards et al. 2019, Wilschut et al. 2019) or adversely stunt plant growth through antagonism (Latz et al. 2016, Schroeder et al. 2020), among other effects.

One of the most important functional groups of soil biota for plant growth is arbuscular mycorrhizal fungi (AMF), which play a key role in P uptake. P deficiency has long been known to

92 negatively affect plant growth (Mengel and Kirkby 1982). Although P is present in the biosphere at high concentrations, plants primarily directly absorb orthophosphate (Pi). Pi is a form scarce in soil (Raghothama 1999) and also essential for microbial growth (Richardson and Simpson 2011, Zhu et al. 2016). Apart from improving plant P acquisition (Collins and Foster 2009), AMF may also contribute to defence against pathogens (St-Arnaud and Vujanovic 2007) and increased abiotic stress tolerance (Augé et al. 2015). The effect of AMF on plant growth can vary widely depending on both plant taxa (Koch et al. 2017) and whether or not associating AMF are native or exotic (Klironomos 2003). In terms of plant P uptake, different species of AMF likewise differ in their efficiency (Miransari et al. 2009). When faced with multiple abiotic stress factors, it is known that plants can benefit in terms of increased biomass and mineral nutrition when grown in soil inoculated with multiple species of AMF compared with soil containing only a single species of AMF (Higo et al. 2018, Crossay et al. 2019, Crossay et al. 2020).

Knowing 1) how the soil legacy of C. scoparius affects plants with differing natural traits (Chapter 2), and 2) the change in fungal community composition entwined in the soil legacy of C. scoparius (Chapter 3), I aimed to examine whether the predominantly positive soil legacy of C. scoparius could be explained in terms of fungal communities.

I hypothesised that:

• One or several changes in fungal diversity induced by C. scoparius (Chapter 3) drives the mostly positive soil legacy effect of C. scoparius (Chapter 2).

• One or several changes in fungal proportional abundance induced by C. scoparius drives the mostly positive soil legacy effect of C. scoparius.

Methods

The starting point relating to my greenhouse data was my raw data on plant biomass and shoot % N and shoot % P which I obtained in Chapter 2, excluding plants found dead at harvest (3.8% of total).

I chose to focus on the fungal diversities within my individual soil eDNA samples (n = 431) as opposed to my pooled eDNA samples, as my pooled eDNA samples only deal with a subset of the 18 plots used in my natural survey. I used R version 3.5.0 (Team 2013) for creating graphs and conducting analyses. All fungal diversity and proportional abundance estimates (e.g., Ascomycota rarefied richness) were based on the same evenly rarefied OTU matrices which I used in Chapter 3 and the subsample size for rarefying my community was set to the minimum number of sequences in any sample. When examining the composition of my samples according to fungal taxon and functional guild, I first generated randomly rarefied community versions of my dataset via the rrarefy function in “vegan” (Oksanen et al. 2013). This process was iterated 250 times before calculating mean rarefied richness and proportional abundances for each fungal taxon and functional guild across all iterations. The rrarefy function was here appropriate so that community metrics are not biased by the size of the sample.

To quantify the effects of the soil legacy of C. scoparius on plant biomass and nutrient composition according to plant species traits, I used linear mixed-effect models via the R package “lme4 (v1.121)” (Bates et al. 2014) to account for plant species and plot. I performed model simplification based on update in the “lme4 (v1.1)” package to narrow down which response variables (if any) best predicted changes in plant biomass, shoot % N and shoot % P. Following the procedure described in Crawley (2002), third-order interactions (or higher) were only considered significant when P < 0.01, while main effects and second-order interactions were considered significant at P < 0.05. All tested continuous variables were scaled when performing mixed-effect models. In order to avoid over-parameterisation of models, I undertook two consecutive steps regarding mixed- effect modelling:

1) What are the drivers of plant growth?

Setting plant species and soil sampling plot as random effects, I tested the effect of all estimates of fungal rarefied richness and proportional abundance per plot for those fungal groups which had shown a significant (P < 0.05) response to C. scoparius at the plot scale in Chapter 3 (Table 2; Chapter 3) against shoot biomass, shoot % N and shoot % P (from Chapter 2).

2) Do the identified drivers of plant growth vary according to plant functional traits?

I then tested whether the explanatory variables obtained in the above step could “fit into” the original model used in my greenhouse experiment. In this original model, I found a significant three-way interaction for shoot biomass between C. scoparius coverage, plant origin, and legume status. Plant species and soil sampling plot were set as random effects while plant origin (i.e., ‘Native’ or ‘Exotic’) and legume status (i.e., ‘Fabaceae’ or ‘Non-Fabaceae’) were set as fixed effects.

Results

What are the drivers of plant growth?

Several explanatory variables were retained in the best model for shoot biomass, shoot % N and shoot % P (Table 1 relating to fungal richness; Table 2 relating to fungal proportional abundance). Distance to C. scoparius and both the proportional abundance and rarefied richness of Glomeromycotina (i.e., AMF) had an effect (P < 0.05) on shoot biomass whereas the rarefied richness of plant pathogens and saprotrophs both affected shoot % N and shoot % P. There was a positive correlation between AMF rarefied richness, AMF proportional abundance and an indicator for C. scoparius coverage (Appendix D1), although it was noted that one of the 18 soil plots had a very high AMF proportional abundance compared to the rest.

Do the identified drivers of plant growth vary according to plant functional traits? – Shoot biomass

Once the chosen explanatory variables (Tables 1 & 2) were added to a model which included plant origin and legume status as fixed effects, a three-way significant interaction was found for shoot biomass between AMF rarefied richness, plant origin, and legume status (t = -4.482; P < 0.0001; Table 3). Although increased AMF richness caused an overall increase in shoot biomass regardless of plant taxonomy or plant functional traits, AMF richness most notably increased the shoot biomass of the exotic Fabaceae Trifolium repens and Ulex europaeus (Figure 1).

Do the identified drivers of plant growth vary according to plant functional traits? – Shoot % N and % P

A three-way significant interaction was found for shoot % N between saprotroph richness, plant origin and legume status (t = 2.841; P = 0.0045; Table 3). When looking at each plant species individually, no discernible effect of saprotroph richness was observed on shoot % N (Appendix D2).

No three-way interaction was observed for shoot % P, although a two-way interaction was observed for shoot % P between plant pathogen richness and legume status (t = 2.411; P = 0.0159; Table 3) as well as saprotroph richness and legume status (t = -2.209; P = 0.0272; Table 3). When looking at each plant species individually, there was no obvious effect of saprotroph or pathogen richness on shoot % P (Appendix D3).

Table 1. Linear mixed-effect model P value extimates and effect sizes for all species rarefied richness predictors and reponse variables which previously showed a significant response to C. scoparius at the plot-scale. Plant species and soil sampling plot were both set as random effects. “.” indicates term dropped during model simplification. Distance to C. scoparius was measured as distance from the extracted soil core to the closest mature C. scoparius (determined by the presence of flowers or seedpods).

Response Predictors (richness) Effect size P Shoot All fungi . . mass (g) Basidiomycota . . Glomeromycotina 0.40 < 0.0001 Chytridiomycotina . . Plant pathogens . . Saproptrophs . . Distance from broom . .

%N (shoot) All fungi . . Basidiomycota . . Glomeromycotina . . Chytridiomycotina . . Plant pathogens -0.17 0.0217 Saproptrophs 0.23 0.0018 Distance from broom . .

%P (shoot) All fungi . . Basidiomycota . . Glomeromycotina . . Chytridiomycotina . . Plant pathogens -0.42 0.0342 Saproptrophs 0.45 0.0250 Distance from broom . .

Table 2. Linear mixed-effect model P value extimates and effect sizes for all species proportional abundance predictors and reponse variables which previously showed a significant response to C. scoparius at the plot-scale. Plant species and soil sampling plot were both set as random effects. “.” indicates term dropped during model simplification. Distance to C. scoparius was measured as distance from the extracted soil core to the closest mature C. scoparius (determined by the presence of flowers or seedpods). Prop. abund. = proportional abundance.

Response Predictors (prop. abund.) Effect size P Shoot Glomeromycotina 0.30 < 0.0001 mass (g) Chytridiomycotina . . Plant pathogens . . Distance from broom -0.19 0.0031

%N (shoot) Glomeromycotina . . Chytridiomycotina . . Plant pathogens . . Distance from broom . .

%P (shoot) Glomeromycotina . . Chytridiomycotina . . Plant pathogens . . Distance from broom . .

Table 3. Linear mixed-effect model P value, t value and parameter estimates for shoot biomass, shoot % N and shoot % P. All predictors retained in the best model are reported, including terms marginal to significant interactions. All used explanatory variables are based of Tables 1 & 2. The explanatory variables for shoot biomass in the initial full model were AMF rarefied richness, AMF proportional abundance and distance to closest mature C. scoparius along with plant origin (‘Native’, TRUE/FALSE) and Fabaceae (TRUE/FALSE) status as fixed effects and with plant species and soil sampling plot as random effects. The explanatory variables for both shoot % N and shoot % P in the initial full model were plant pathogen rarefied richness and saprotroph rarefied richness along with plant origin and Fabaceae status as fixed effects (and plant species and soil sampling plot as random effects).

Shoot biomass (g) P t Parameter estimates AMF richness < 0.0001 0.269 0.025 Distance to closest mature broom 0.0494 -1.965 -0.131 Fabaceae 0.0352 3.394 1.063 Native 0.1860 0.966 0.302 AMF richness × Fabaceae < 0.0001 7.284 0.771 AMF richness × Native 0.0643 1.825 0.191 Fabaceae × Native 0.0072 -2.684 -1.188 AMF richness × Fabaceae × Native < 0.0001 -4.482 -0.669

%N (shoot) P t Parameter estimates Saprotroph richness 0.0342 2.840 0.228 Fabaceae < 0.0001 4.325 1.422 Native 0.1186 -0.472 -0.155 Saprotroph richness × Fabaceae 0.9408 -1.966 -0.216 Saprotroph richness × Native 0.5225 -2.388 -0.249 Fabaceae × Native 0.3671 -0.901 -0.420 Saprotroph richness × Fabaceae × Native 0.0045 2.841 0.439

%P (shoot) P t Parameter estimates Plant pathogen richness 0.0344 -3.056 -0.716 Saprotroph richness 0.0245 3.063 0.716 Fabaceae 0.7336 -1.730 -0.534 Native 0.3764 -0.835 -0.256 Plant pathogen richness × Fabaceae 0.0159 2.411 0.623 Saprotroph richness × Fabaceae 0.0272 -2.209 -0.572 Fabaceae × Native 0.0355 2.102 0.921

Figure 1. Standardised and centred aboveground dry biomass (g) over rarefied richness of AMF (i.e., Glomeromycotina). Size of points is scaled according to the square root of C. scoparius coverage. Lines were obtained from the created linear mixed-effect model via lme4’s predict function.

100

Discussion

The results of this study suggest that increased AMF richness induced by the soil legacy of C. scoparius favours the growth of Fabaceae, particularly exotic Fabaceae, compared with non-N- fixing native plants. This finding matches the results of our field experiment, which likewise showed that the association of C. scoparius with soil microorganisms was most beneficial for exotic legumes compared to native legumes (Allen et al. (2020); Appendix E). Once AMF rarefied richness was included in my models, there was no residual effect of C. scoparius coverage on shoot biomass, inferring that AMF rarefied richness was sufficient to explain effects on shoot biomass. I accept my hypothesis that a change in fungal diversity induced by C. scoparius drives the mostly positive soil legacy effect of C. scoparius and can specify that AMF richness (as opposed to AMF proportional abundance) is here responsible.

Effect of AMF richness on plant growth

Although it was statistically supported that exotic Fabaceae benefited the most from increased AMF rarefied richness, this pattern was mostly attributed to Trifolium repens and Ulex europaeus and was least apparent for Lupinus arboreus (Figure 1). As L. arboreus is considered non-mycorrhizal (Oba et al. 2001), L. arboreus may to some extent serve as “the exception that proves the rule”, considering that L. arboreus could not benefit from increased AMF richness due to not forming AMF associations.

A straightforward explanation for the increase in shoot biomass observed for Fabaceae would be that due to the lack of N-fixation by rhizobia, non-Fabaceae are more N-limited than Fabaceae and therefore cannot effectively profit from increased P acquisition from AMF as N remains a limiting factor. In more detail, I already know from the results of my greenhouse experiment that exotic Fabaceae in particular were putatively P limited (Koerselman and Meuleman 1996), whereas most non-Fabaceae were putatively N limited (Figure 6; Chapter 2). As such, having an increased P availability will not necessarily benefit the growth of N-limited plants (i.e., non- Fabaceae), whereas having an increased P availability is likely to benefit the growth of N-fixing plants which are not N-limited (i.e., Fabaceae). It is also worth noting that non-mycorrhizal L. arboreus was markedly more putatively P limited than other exotic Fabaceae (Figure 6; Chapter 2), which could be expected due to L. arboreus being unable to derive P from AMF (although see Wang et al. (2019) regarding the ability of some Lupinus sp. to instead form cluster roots to increase P uptake).

It is nonetheless uncertain as to why exotic Fabaceae benefit more from increased AMF richness than native Fabaceae. I know from the results of my natural survey (Chapter 3) that 1) C. scoparius

101 is possibly spreading alongside belowground mutualists, pathogens and commensals (Nuñez and Dickie 2014), while simultaneously forming cosmopolitan associations with widespread AMF communities across plots. I know from our field experiment that 2) the association of C. scoparius with soil microorganisms was most beneficial for exotic legumes compared to native legumes (Allen et al. (2020); Appendix E). Although the response of C. scoparius to AMF richness was not as strong as Trifolium repens or Ulex europaeus, considering points 1) and 2), it may be proposed that the exotic legumes in my experiment were more amenable to forming both novel and cosmopolitan AMF associations than native legumes and can therefore benefit more from available AMF symbioses. Although plant-AMF mutualisms are more promiscuous than plant-bacterial mutualisms (Klironomos et al. 2000), it has already been proposed that legumes native to New Zealand form different associations with N-fixing rhizobia compared with exotic Cytisus scoparius and Ulex europaeus (Weir et al. 2004). It may be the case that native and exotic Fabaceae likewise associate selectively with AMF, either in terms of more specific associations or an increased number of associations. Though not explicit to AMF, there is growing evidence that exotic plants integrate into broader, more generalist association networks (Stouffer et al. 2014, Rodríguez- Echeverría and Traveset 2015, Emer et al. 2016), although there are exceptions to this pattern as native grasses, for instance, can form more associations with a given diversity of AMF compared with non-native grasses (Jordan et al. 2012).

Different species of AMF have been known to lead to different degrees of plant P uptake (Miransari et al. 2009). The fact that AMF rarefied richness was a stronger predictor of plant biomass as opposed to AMF proportional abundance might indicate that a higher AMF richness correlates with an increased likelihood that one or several associating AMF confer relatively more P to plants. Plants can discriminate and reward the best AMF partners with more carbohydrates (Kiers et al. 2011). In turn, a plant’s AMF can enforce cooperation by increasing nutrient transfer only to those roots providing more carbohydrates (Helgason et al. 2002, Kiers et al. 2011). A more speculative interpretation in a similar vein with Jansa et al. (2008) regarding plant growth and AMF richness would be that AMF are forced into inter-specific competition as a plant will preferentially associate itself with the species of AMF which supplies the greatest amount of benefits. A plant with a high diversity of AMF would receive P “at a bargain price”, not only because the plant initially has a high number of AMF species which can supply P, but also because associating AMF species now need to compete against each other when “setting the price of nutrient exchange”. Such belowground economics may however be biased toward plant growth in greenhouse experiments, as having a single plant per pot forces any existing AMF to rely on that particular plant for mutualism.

102

For a given plant species, soil obtained from closely related plants generally has a more negative effect on plant growth than soil obtained from distantly related plants (Kempel et al. 2018). Negative plant-soil feedbacks occur in part when pathogens accumulate in the rhizosphere of plant species (Kulmatiski et al. 2008). This accumulation of plant pathogens alongside C. scoparius coverage (both in terms of richness and proportional abundance) has been observed in my natural survey (Chapter 3). Regarding why C. scoparius did not show as notable an increase in shoot biomass over AMF richness compared to T. repens and U. europaeus, it is likely that negative feedback of plant pathogens more specific to C. scoparius counteracted the effect of increased AMF richness.

Experimental design considerations

Most experimental design considerations in this chapter overlap those in Chapters 2 & 3. Although I studied how soil chemical and soil biological attributes correlate with plant growth, the physical aspect of soil legacy has not been considered, despite having an important impact on plant development (Van der Putten et al. 2013). In the design of my greenhouse experiment, I mixed the soils I collected across a C. scoparius density gradient at a 1:1 volume ratio with washed river sand. This mixing was done to minimize the quantity of soil required (and thereby any disruption caused to the permanent sampling plots) as well as to help standardize soil porosity and generally reduce physical disparities between soils. Although obtained from the same 18 plots, I also cannot claim that the soil community to which plants were exposed to in my greenhouse experiment is the exact same soil community in my natural survey. However, as I examined changes in microbial composition across a C. scoparius density gradient (as opposed to a “Control and Effect” style experiment), I can be reasonably confident that the effects of both soil legacy and soil community composition overlap each other.

Conclusions and applications

Although soil legacy studies are increasingly popular (Klironomos 2003, Edwards et al. 2019, Wilschut et al. 2019), it is fairly rare that a soil legacy study traces growth responses back to fungal community composition. Though I previously thought that the removal of C. scoparius might benefit the growth of native plants due to the generally positive soil legacy of C. scoparius, I now see that AMF richness is integral to the legacy of C. scoparius and may benefit P-limited exotic Fabaceae to a greater extent than N-limited non-Fabaceae as well as native Fabaceae. I add that growth benefits to P-limited exotic Fabaceae are contingent on the ability of the exotic Fabaceae to form AMF associations, as non-mycorrhizal plants (i.e., L. arboreus) did not profit from an increase in AMF richness. A recent study has shown how plant growth may increase when a plant is co-inoculated with both AMF and certain bacteria (Bourles et al. 2020). A future research avenue would be to further examine how known differences in the rhizobial associations between native

103 and exotic legumes (Weir et al. 2004) might explain why exotic Fabaceae differ in their growth response to AMF richness compared to native Fabaceae.

104

Chapter 6: General discussion

Key findings

My PhD thesis represents a body of research work which combined a well-established soil legacy approach with more modern eDNA metabarcoding, enabling me to observe how a high diversity of fungi, particularly arbuscular mycorrhizal fungi, underlies the invasion success of C. scoparius. Although fungal diversity in soil under C. scoparius is higher than in grassland uninvaded by C. scoparius, C. scoparius invasion results in increased homogenisation of certain fungal groups within the overall soil fungal community. Increased arbuscular mycorrhizal richness, which is found in these more homogenised soil communities, is partly responsible for the generally positive soil legacy of C. scoparius, especially for exotic Fabaceae which can probably benefit more from AMF- facilitated P enrichment due to their ability to first fix required N for growth. The benefit that exotic Fabaceae derive from C. scoparius can be observed both in greenhouse studies as well as in the presence of live C. scoparius. I present the pitfalls and benefits of eDNA pooling, show a fungal taxon-wide bias in the proportional abundance of fungi in pooled samples, demonstrate how rarer fungi remain increasingly unaccounted for with increased degrees of pooling, yet also show how pooling may benefit researchers who wish to study larger-scale processes.

Considerations for eDNA community studies

There has generally been a correlation between the biodiversity of groups of directly or indirectly interacting organisms (Gaston 2000, Scherber et al. 2010, Peng et al. 2019). Having sampled from plots where C. scoparius formed near-monocultures, the most counterintuitive result in my thesis was the discovery that C. scoparius increased fungal species richness compared to uninvaded grassland. Soil biological responses to plant invasion do not evidently need to result in decreases in species richness as has been documented for native plants (D'Antonio and Flory 2017, Fahey et al. 2018), arthropods (Andersen et al. 2019, Jesse et al. 2020), birds (Grzędzicka and Reif 2020), small and large mammals (Ceradini and Chalfoun 2017, Dumalisile and Somers 2017), and amphibians (Nunes et al. 2019), among other taxa. The field of invasion ecology has generally held an aboveground focus (Dickie et al. 2017a), although more belowground studies have been conducted since the advent of eDNA metabarcoding. Knowledge of how fungal communities change across the density gradient of a plant invasion has not only permitted me to describe the consequences of a plant invasion, but also identify possible belowground enablers of plant invasion. Whereas the majority of invasion ecology studies focus on either the cause (e.g., habitat fragmentation) or effect (e.g., reduced richness) of a plant invasion, eDNA metabarcoding presents

105 a novel opportunity to consider both causes and effects simultaneously, while also obtaining a general overview of how species are distributed in certain sites.

Although eDNA metabarcoding has been implemented to assess New Zealand’s terrestrial biodiversity (Holdaway et al. 2017), metabarcoding is subject to several biases (Bulman et al. 2018, Makiola et al. 2019b), which has led to calls for eDNA metabarcoding studies requiring robust experimental designs to draw sound ecological conclusions (Dickie et al. 2018, Zinger et al. 2019). I demonstrate in my eDNA pooling experiment (Chapter 4) that biases surrounding eDNA metabarcoding studies extend to eDNA sample pooling, which modifies both the observed rarefied richness and proportional abundance of fungi.

As analysing eDNA communities is a rapidly evolving and diverse field (Taberlet et al. 2018, Allwood et al. 2020), there are always variations in methodology to consider when undertaking eDNA studies. Although the soil extraction kit I used in my eDNA natural experiment (Chapter 3) and my eDNA pooling experiment (Chapter 4) was the same recommended by Lear et al. (2018), there still remains some uncertainty regarding the best available method to process soil samples (Hermans et al. 2018). It is known that eDNA metabarcoding studies are likewise affected by substrate selection (Koziol et al. 2019) and I can be reasonably confident that using deeper soil samples (deeper than 150-200 mm) would have impacted observed fungal communities (Schlatter et al. 2018, Sosa-Hernández et al. 2018). However, as exotic plants often have shallower roots than native species (Upton et al. 2020), studying the effect of C. scoparius on topsoil fungal communities does give a more reliable picture of changes in fungal communities induced by C. scoparius. Topsoil is also most relevant for understanding effects on seedlings of other species, as all seedlings start with their roots in the topsoil. A benefit of obtaining soil samples from permanent sampling plots laid out according to Hurst and Allen (1993) is that longer-term changes can be measured, which is relatively rare in soil community literature (although see Sielaff et al. (2018) or Song et al. (2020)).

As recommended by Dickie et al. (2018), I used both negative and positive controls in my experimental design to account for possible sample contamination (which proved low) and to set aside any DNA “naturally” found in extraction kits (Toole et al. 2019). As I performed my PCRs in duplicate for both my natural survey (Chapter 3) and my eDNA pooling experiment (Chapter 4), I can be more confident that my results are more reproducible (Bautista-de los Santos et al. 2016). A very important consideration for eDNA studies is the choice of primers. It was my initial intention to use the ITS7o primer, as it enables better identification of arbuscular mycorrhizal fungal communities (Kohout et al. 2014), which proved important throughout my results. However, an initial unsuccessful metabarcoded library submission with the ITS7o primer prompted me to revert to the more widely used ITS7 primer (Ihrmark et al. 2012). Other colleagues had reported similar issues, and it is probable that faulty DNA normalization and purification

106 plates had led to my initial unsuccessful eDNA library submission as opposed to an issue with the ITS7o primer itself. The clarity of my results would have nonetheless benefited from using the ITS7o primer instead of the ITS7 primer and it is probable that the diversity and proportional abundance estimates I obtained for arbuscular mycorrhizal fungi are underestimated.

It is unfortunately common that eDNA surveys give few details on the spatial arrangement and sampling design implemented (Dickie et al. 2018). Using a systematic method for sample collection (Hurst and Allen 1993), I was able to simultaneously 1) ensure the reproducibility of my results, 2) enable even comparisons between sampling plots, and 3) present my results via a measurement of soil community heterogeneity (i.e., true beta diversity (Whittaker 1970)), which is more easily understood by the public.

Lessons for soil legacy studies

The identity and composition of soil biota which may underlie plant growth responses or nutrient acquisition are rarely accounted for, even though soil microbiota can reduce plant growth through antagonism (Latz et al. 2016, Schroeder et al. 2020) and may determine plant nutrient availability (Edwards et al. 2019, Wilschut et al. 2019), among other effects. There have been calls to implement microbiome sequencing techniques in order to develop more predictive plant-soil feedback frameworks in certain ecosystems (Singh and Meyer 2020) and more generally understand how exotic plants can spread (Collins et al. 2019, Ramirez et al. 2019). My results in Chapter 5 show that knowing the fungal community inherent to the soil legacy of C. scoparius enabled a clearer understanding of the belowground processes which aid in the invasion of exotic Fabaceae, or more specifically, exotic Fabaceae which can form associations with AMF.

Soil legacy studies have moved from straightforward “Control and Effect” style experiments (Bever 1994) to more complex investigations which look at how both chemical and biological changes to soil composition induced by different densities of an invasive species alter the growth of plants with varying natural traits (Chapter 2; Chapter 5). Soil legacy studies incorporating soil communities are becoming increasingly popular (Detheridge et al. 2016, Pickett et al. 2019). Soil legacy responses can result in more dramatic changes to soil fungi compared to soil bacteria (Heinen et al. 2020), partly as fungi tend to be comparatively more spatially heterogeneously distributed in soil (Manter et al. 2010) (see however Collins et al. (2016) regarding how a plant can modify bacterial communities to a greater extent than fungal communities). Having studied fungal communities, a possible future research avenue would be to explore the communities of other soil micro-organisms, such as oomycetes (Cacciola and Gullino 2019), viruses (Sutela et al. 2019) and tardigrades (Bryndová et al. 2020), which may affect plant growth. Although every soil micro- organism can be an important determinant of soil legacy, considering which group or groups of soil micro-organisms could have impacted soil legacy studies could well increase our

107 understanding of plant-soil feedbacks. There have already been examples in the literature where both fungi and bacteria have been shown to be inherent to a plant’s soil legacy (Bourles et al. 2020, Pan et al. 2020, Saia et al. 2020a), and it is likely that studying these soil communities will shed more light on belowground processes which contribute to the spread of exotic plants (Waller et al. 2020).

Considerations for the conservation of New Zealand’s grassland

The rapid growth of C. scoparius along with the formation of seedbanks make C. scoparius extraordinarily resistant to eradication, even after applying herbicide (Tran et al. 2016, Haubensak et al. 2020). However, ecological restoration with native vegetation could suppress C. scoparius, which is shade intolerant (Watt et al. 2003b, Burrows et al. 2015). Plant-soil feedbacks are important when considering ecological restoration projects (Yelenik and Levine 2011, Wubs et al. 2016), as new generations of seedlings might benefit or suffer from changes imposed by a previous plant on its surrounding soil. Monitoring changes in fungal communities is likewise important in restoration projects (Yan et al. 2018) as there can be a strong link between fungal diversity and soil and plant properties (Tedersoo et al. 2016, Yang et al. 2017). An ecological restoration project involving the removal of C. scoparius is something of a double-edged sword both in terms of species richness and soil legacy.

In terms of species richness, removing near-monocultures of C. scoparius would obviously increase plant diversity at least in the short term, yet as I found a greater fungal richness in soil invaded by C. scoparius compared to uninvaded grassland with a higher plant species richness (Chapter 3), C. scoparius removal may likely result in a decrease in fungal diversity integral to New Zealand’s conservation efforts (Holdaway et al. 2017, de Lange et al. 2018, Dickie et al. 2020). Moreover, as I unexpectedly found a greater proportion of rare fungi in plots with high C. scoparius coverage, removing large C. scoparius invasions might possibly endanger rarer local soil-dwelling fungi (although see Dickie et al. (2020) concerning how rare wood inhabiting fungi are not local). However, it is important to consider the relatively small scale at which I undertook my natural survey. Although the increased productivity induced by C. scoparius may putatively enable more fungal OTUs to exist near C. scoparius (Chapter 3), I also know that C. scoparius decreases heterogeneity of some fungal groups, and it is probable that broader scale eDNA sampling of grassland uninvaded by C. scoparius will reveal a greater fungal diversity compared to the more homogeneous fungal communities near C. scoparius.

In terms of soil legacy, soil obtained after removing live C. scoparius does increase overall plant growth in a greenhouse environment (Chapter 2), although exotic Fabaceae benefited more than native Fabaceae in both my greenhouse experiment and our field experiment (Allen et al. (2020); Appendix E). Cytisus scoparius frequently grows in New Zealand’s forestry plantations where it is

108 considered detrimental to the country’s pine industry, particularly for Pinus radiata (Tran et al. 2016). Although generally regarded as a serious issue in early stages of pine development (Carter et al. 2019a), it is possible that C. scoparius might actually benefit older pine stands partly through N enrichment. This benefit is apparent for young P. radiata in a greenhouse environment, which show increased shoot % N and increased shoot N:P ratio over C. scoparius coverage (Chapter 2). My results are, however, based on soil free from live C. scoparius, which may be difficult to achieve in the field once C. scoparius is established (Haubensak et al. 2020). One innovative way to restore ecosystems following an invasion by non-native Fabaceae has been to allow the Fabaceae to “run its course”, which then permits native flora to benefit from the Fabaceae’s N-fixation, as has been implemented in Hinewai Reserve (New Zealand) for Ulex europaeus (Wilson et al. 2017). Cytisus scoparius can in some sites be regarded as a nurse crop for recovery of indigenous woody vegetation (Burrows et al. 2015). It can often be only a matter of time before plant pathogen-induced plant- soil feedback restricts the growth of older C. scoparius invasions, allowing other plants to benefit from increased soil N availability (Burrows et al. 2015).

More broadly in terms of plant invasions, prevention is the best cure (Hulme 2020). Five biocontrol agents have already been introduced to New Zealand to combat the spread of C. scoparius (Syrett et al. 1999, Syrett et al. 2007, Paynter et al. 2012), yet the shrub has become so commonplace in New Zealand that it now commonly regarded as part of the country’s ‘natural’ landscape. For example, C. scoparius can be seen in promotional pictures for New Zealand’s tourism industry (www.tourismnewzealand.com; accessed 16/06/2020). Plants from the taxonomic family of C. scoparius (Fabaceae) already have a long history of being invasive in areas which have undergone disturbance (Figueiral and Bettencourt 2004). Reinhart et al. (2017) observed that plant invasiveness is generally not associated with mycorrhizal responsiveness, yet acknowledge that mycorrhizal responsiveness can contribute to invasiveness in certain species. Cytisus scoparius and other leguminous shrubs (e.g., Ulex europaeus) may be the exception rather than the norm in terms of using AMF as a means to competitively increase growth. Considering the dependence of exotic Fabaceae on belowground symbionts, there is a silver lining in that plants can take several years to cultivate typical microbial communities and it likewise takes several years before plant-soil feedbacks come into full effect (Kulmatiski and Beard 2011). Now having a clearer image of the underlying mechanisms of C. scoparius, conservation effort should be prioritised toward restricting the spread of exotic Fabaceae which exist in relatively smaller ranges of New Zealand, before the invasion of these Fabaceae gains too much momentum. A list of such exotic Fabaceae along with their distribution throughout New Zealand can be found in Howell and Terry (2016) and include Dipogon lignosus and Genista monspessulana.

109

Future research interest: Generalist interactions between C. scoparius and belowground symbionts

Plants can share specific fungal symbionts at the level of populations or species (Dickie et al. 2017b). In a similar way as interaction networks have been applied, for example, in pollination biology to study which plants interact with which insects (Traveset et al. 2013), interaction networks may also be applied to study close associations between fungi and plants (Dickie et al. 2017a, Zenni et al. 2017, Xiao et al. 2018). Although some plants have been shown to be successful invaders via a single fungal symbiont (Hayward et al. 2015), having multiple fungal symbionts might also account for a plant’s improved growth (Higo et al. 2018, Crossay et al. 2020). Having analysed soil samples as opposed to root samples with associating fungi, I cannot be certain that the increased richness of AMF I found in soil with higher C. scoparius density interacts with C. scoparius. Nonetheless, my results do indicate that apart from C. scoparius increasing the richness of surrounding fungi (including AMF), C. scoparius may putatively form more interactions with more species of AMF compared to other plants in uninvaded grasslands (Chapter 3).

A future research interest would be to analyse C. scoparius and its belowground symbionts via interaction networks. For each soil sample I used in my natural survey, I have also collected accompanying root samples with associating fungi, which may yield the raw data necessary for network analysis. Given the increases in soil fungal richness linked with higher densities of C. scoparius, I would hypothesize that C. scoparius is generalist in its associations with belowground mutualists to the extent that C. scoparius may act as a figurative “fungal sponge”. Alien tree- ectomycorrhizal communities have been shown to be able to form their own network in novel ranges as well as form new linkages with native interaction networks (Dickie et al. 2017b), which may underlie their invasiveness. I am keen to discover whether non-native legume-arbuscular mycorrhizal networks likewise integrate into native networks.

Parting words

Metabarcoding is an exciting technique which can bring fresh insight to long-studied ecological processes such as invasion (Deiner et al. 2017) and plant-soil feedback (Dierks et al. 2019). Whereas invasion ecology has classically held an aboveground focus, much remains to be said regarding microbial invasions and plants co-invading with belowground mutualists (Dickie et al. 2017a). Sequencing eDNA has allowed us to gain insight into mostly hidden belowground communities, however this process comes hand in hand with challenges in terms of how to set-up experiments and interpret obtained results (Dickie et al. 2018, Lear et al. 2018). Two key themes which underlie my results are the importance of regulation and integration. Regulation of metabarcoding studies is required to accurately draw out reproducible conclusions and integrating soil legacy studies with eDNA data can provide a more comprehensive view of processes underlying plant invasion.

110

References

Acinas, S. G., R. Sarma-Rupavtarm, V. Klepac-Ceraj, and M. F. Polz. 2005. PCR-induced sequence artifacts and bias: insights from comparison of two 16S rRNA clone libraries constructed from the same sample. Applied and Environmental Microbiology 71:8966- 8969. Aldorfová, A., P. Knobová, and Z. Münzbergová. 2020. Plant‐soil feedback contributes to predicting plant invasiveness of 68 alien plant species differing in invasive status. Oikos 00:1-14 Allen, R., P. Williams, and W. Lee. 1995. Seed bank accumulation of broom (Cytisus scoparius) in South Island. Proceedings of the 48th New Zealand Plant Protection Conference:276-280. Allen, W. J., R. Wainer, J. M. Tylianakis, B. I. Barratt, M. R. Shadbolt, L. P. Waller, and I. A. Dickie. 2020. Community‐level direct and indirect impacts of an invasive plant favour exotic over native species. Journal of Ecology 00:1–12 Allwood, J. S., N. Fierer, and R. R. Dunn. 2020. The future of environmental DNA in forensic science. Applied and Environmental Microbiology 86:e01504-19. Altschul, S. F., T. L. Madden, A. A. Schäffer, J. Zhang, Z. Zhang, W. Miller, and D. J. Lipman. 1997. Gapped BLAST and PSI-BLAST: a new generation of protein database search programs. Nucleic Acids Research 25:3389-3402. Andersen, E. M., M. N. Cambrelin, and R. J. Steidl. 2019. Responses of grassland arthropods to an invasion by nonnative grasses. Biological Invasions 21:405-416. Anthony, M., S. Frey, and K. Stinson. 2017. Fungal community homogenization, shift in dominant trophic guild, and appearance of novel taxa with biotic invasion. Ecosphere 8:e01951. Anthony, M., K. Stinson, A. Trautwig, E. Coates-Connor, and S. Frey. 2019. Fungal communities do not recover after removing invasive Alliaria petiolata (garlic mustard). Biological Invasions 21:3085-3099. Aslam, S., A. Tahir, M. F. Aslam, M. W. Alam, A. A. Shedayi, and S. Sadia. 2017. Recent advances in molecular techniques for the identification of phytopathogenic fungi–a mini review. Journal of Plant Interactions 12:493-504. Asner, G. P., R. F. Hughes, P. M. Vitousek, D. E. Knapp, T. Kennedy-Bowdoin, J. Boardman, R. E. Martin, M. Eastwood, and R. O. Green. 2008. Invasive plants transform the three- dimensional structure of rain forests. Proceedings of the National Academy of Sciences 105:4519-4523. Atkins, S. D., and I. M. Clark. 2004. Fungal molecular diagnostics: a mini review. Journal of Applied Genetics 45:3-15. Augé, R. M., H. D. Toler, and A. M. Saxton. 2015. Arbuscular mycorrhizal symbiosis alters stomatal conductance of host plants more under drought than under amply watered conditions: a meta-analysis. Mycorrhiza 25:13-24. Avis, P., S. Branco, Y. Tang, and G. Mueller. 2010. Pooled samples bias fungal community descriptions. Molecular Ecology Resources 10:135-141. Avis, P., G. Mueller, and J. Lussenhop. 2008. Ectomycorrhizal fungal communities in two North American oak forests respond to nitrogen addition. New Phytologist 179:472-483. Bachelot, B., M. Uriarte, J. K. Zimmerman, J. Thompson, J. W. Leff, A. Asiaii, J. Koshner, and K. McGuire. 2016. Long‐lasting effects of land use history on soil fungal communities in second‐growth tropical rain forests. Ecological Applications 26:1881-1895. Badri, D. V., and J. M. Vivanco. 2009. Regulation and function of root exudates. Plant, Cell & Environment 32:666-681. Bahram, M., T. Netherway, F. Hildebrand, K. Pritsch, R. Drenkhan, K. Loit, S. Anslan, P. Bork, and L. Tedersoo. 2020. Plant nutrient‐acquisition strategies drive topsoil microbiome structure and function. New Phytologist. doi.org/10.1111/nph.16598.

111

Bais, H. P., R. Vepachedu, S. Gilroy, R. M. Callaway, and J. M. Vivanco. 2003. Allelopathy and exotic plant invasion: from molecules and genes to species interactions. Science 301:1377- 1380. Bakker, P. A., C. M. Pieterse, R. de Jonge, and R. L. Berendsen. 2018. The soil-borne legacy. Cell 172:1178-1180. Balami, S., L. B. Thapa, and S. K. Jha. 2017. Effect of invasive Ageratina adenophora on species richness and composition of saprotrophic and pathogenic soil fungi. BIOTROPIA-The Southeast Asian Journal of Tropical Biology 24:212-219. Balvanera, P., A. B. Pfisterer, N. Buchmann, J. S. He, T. Nakashizuka, D. Raffaelli, and B. Schmid. 2006. Quantifying the evidence for biodiversity effects on ecosystem functioning and services. Ecology Letters 9:1146-1156. Bardgett, R. D., and W. H. Van Der Putten. 2014. Belowground biodiversity and ecosystem functioning. Nature 515:505-511. Bascand, L., and G. Jowett. 1982. Scrubweed cover of South Island agricultural and pastoral land 2. Plant distribution and managerial problem status. New Zealand Journal of Experimental Agriculture 10:455-492. Bates, D., M. Maechler, B. Bolker, S. Walker, R. H. B. Christensen, H. Singmann, and B. Dai. 2014. Linear mixed-effects models using Eigen and S4. R package version 1. Bautista-de los Santos, Q. M., J. L. Schroeder, O. Blakemore, J. Moses, M. Haffey, W. Sloan, and A. J. Pinto. 2016. The impact of sampling, PCR, and sequencing replication on discerning changes in drinking water bacterial community over diurnal time-scales. Water Research 90:216-224. Beckerman, A. P., and O. L. Petchey. 2009. Infectious food webs. Journal of Animal Ecology 78:493-496. Bellingham, P. 1998. Shrub succession and invasibility in a New Zealand montane grassland. Australian Journal of Ecology 23:562-573. Bellingham, P., and D. Coomes. 2003. Grazing and community structure as determinants of invasion success by Scotch broom in a New Zealand montane shrubland. Diversity and Distributions 9:19-28. Bever, J. D. 1994. Feeback between plants and their soil communities in an old field community. Ecology 75:1965-1977. Bever, J. D. 2002. Negative feedback within a mutualism: host–specific growth of mycorrhizal fungi reduces plant benefit. Proceedings of the Royal Society of London. Series B: Biological Sciences 269:2595-2601. Bever, J. D. 2003. Soil community feedback and the coexistence of competitors: conceptual frameworks and empirical tests. New Phytologist 157:465-473. Bever, J. D., I. A. Dickie, E. Facelli, J. M. Facelli, J. Klironomos, M. Moora, M. C. Rillig, W. D. Stock, M. Tibbett, and M. Zobel. 2010. Rooting theories of plant community ecology in microbial interactions. Trends in Ecology & Evolution 25:468-478. Bever, J. D., K. M. Westover, and J. Antonovics. 1997. Incorporating the soil community into plant population dynamics: the utility of the feedback approach. Journal of Ecology 85:561-573. Bode, R. F., S. Grove, and N. Krueger. 2019. Limits to biocontrol: the effects of urbanization and elevation on Bruchidius villosus and Exapion fuscirostre—two biological control agents of Cytisus scoparius. Biological Invasions 21:1021-1031. Bode, R. F., and R. Tong. 2018. Pollinators exert positive selection on flower size on urban, but not on rural Scotch broom (Cytisus scoparius L. Link). Journal of Plant Ecology 11:493-501. Bogar, L. M., I. A. Dickie, and P. G. Kennedy. 2015. Testing the co‐invasion hypothesis: ectomycorrhizal fungal communities on Alnus glutinosa and Salix fragilis in New Zealand. Diversity and Distributions 21:268-278. Bond, E. M., and J. M. Chase. 2002. Biodiversity and ecosystem functioning at local and regional spatial scales. Ecology Letters 5:467-470. Bossard, C. C. 1993. Seed germination in the exotic shrub Cytisus scoparius (Scotch broom) in California. Madrono 40:47-61.

112

Bourles, A., L. Guentas, C. Charvis, S. Gensous, C. Majorel, T. Crossay, Y. Cavaloc, V. Burtet- Sarramegna, P. Jourand, and H. Amir. 2020. Co-inoculation with a bacterium and arbuscular mycorrhizal fungi improves root colonization, plant mineral nutrition, and plant growth of a Cyperaceae plant in an ultramafic soil. Mycorrhiza: 30(1):121-131 Bradstreet, R. B. 1954. Kjeldahl method for organic nitrogen. Analytical Chemistry 26:185-187. Branco, S., T. D. Bruns, and I. Singleton. 2013. Fungi at a small scale: spatial zonation of fungal assemblages around single trees. PLoS ONE 8(10): e78295. Brandes, U., B. B. Furevik, L. R. Nielsen, E. D. Kjær, L. Rosef, and S. Fjellheim. 2019. Introduction history and population genetics of intracontinental scotch broom (Cytisus scoparius) invasion. Diversity and Distributions 25:1773-1786. Broadbent, A. A., K. H. Orwin, D. A. Peltzer, I. A. Dickie, N. W. Mason, N. J. Ostle, and C. J. Stevens. 2017. Invasive N-fixer impacts on litter decomposition driven by changes to soil properties not litter quality. Ecosystems 20:1151-1163. Broeckling, C. D., A. K. Broz, J. Bergelson, D. K. Manter, and J. M. Vivanco. 2008. Root exudates regulate soil fungal community composition and diversity. Applied and Environmental Microbiology 74:738-744. Broz, A. K., D. K. Manter, and J. M. Vivanco. 2007. Soil fungal abundance and diversity: another victim of the invasive plant Centaurea maculosa. The ISME Journal 1:763-765. Brundrett, M. 1991. Mycorrhizas in natural ecosystems. Pp. 171-313 In Advances in ecological research. Elsevier. Brundrett, M. C. 2009. Mycorrhizal associations and other means of nutrition of vascular plants: understanding the global diversity of host plants by resolving conflicting information and developing reliable means of diagnosis. Plant and Soil 320:37-77. Bryndová, M., D. Stec, R. O. Schill, Ł. Michalczyk, and M. Devetter. 2020. Dietary preferences and diet effects on life-history traits of tardigrades. Zoological Journal of the Linnean Society 188:865-877. Bulman, S., R. McDougal, K. Hill, and G. Lear. 2018. Opportunities and limitations for DNA metabarcoding in Australasian plant-pathogen biosecurity. Australasian Plant Pathology 47:467-474. Burdon, J. 1993. The structure of pathogen populations in natural plant communities. Annual Review of Phytopathology 31:305-323. Burdon, J. J. 1987. Diseases and plant population biology. Cambridge University Press. UK. Burke, D. J., S. R. Carrino-Kyker, A. Hoke, S. Cassidy, L. Bialic-Murphy, and S. Kalisz. 2019. Deer and invasive plant removal alters mycorrhizal fungal communities and soil chemistry: Evidence from a long-term field experiment. Soil Biology and Biochemistry 128:13-21. Burrows, L., E. Cieraad, and N. Head. 2015. Scotch broom facilitates indigenous tree and shrub germination and establishment in dryland New Zealand. New Zealand Journal of Ecology 39:61-70. Cacciola, S. O., and M. L. Gullino. 2019. Emerging and re-emerging fungus and oomycete soil- borne plant diseases in Italy. Phytopathologia Mediterranea 58:451-472. Calderón‐Sanou, I., T. Münkemüller, F. Boyer, L. Zinger, and W. Thuiller. 2019. From environmental DNA sequences to ecological conclusions: How strong is the influence of methodological choices? Journal of Biogeography 00:1–14 Caldwell, B. A. 2006. Effects of invasive scotch broom on soil properties in a Pacific coastal prairie soil. Applied Soil Ecology 32:149-152. Callaway, R., B. Newingham, C. A. Zabinski, and B. E. Mahall. 2001. Compensatory growth and competitive ability of an invasive weed are enhanced by soil fungi and native neighbours. Ecology Letters 4:429-433. Callaway, R. M., G. C. Thelen, A. Rodriguez, and W. E. Holben. 2004. Soil biota and exotic plant invasion. Nature 427:731-733. Caporaso, J. G., C. L. Lauber, W. A. Walters, D. Berg-Lyons, J. Huntley, N. Fierer, S. M. Owens, J. Betley, L. Fraser, and M. Bauer. 2012. Ultra-high-throughput microbial community analysis on the Illumina HiSeq and MiSeq platforms. The ISME Journal 6:1621-1624.

113

Caporaso, J. G., C. L. Lauber, W. A. Walters, D. Berg-Lyons, C. A. Lozupone, P. J. Turnbaugh, N. Fierer, and R. Knight. 2011. Global patterns of 16S rRNA diversity at a depth of millions of sequences per sample. Proceedings of the National Academy of Sciences 108:4516-4522. Cappai, C., A. R. Kemanian, A. Lagomarsino, P. P. Roggero, R. Lai, A. E. Agnelli, and G. Seddaiu. 2017. Small-scale spatial variation of soil organic matter pools generated by cork oak trees in Mediterranean agro-silvo-pastoral systems. Geoderma 304:59-67. Carini, P., M. Delgado-Baquerizo, E.-L. S. Hinckley, H. Holland‐Moritz, T. E. Brewer, G. Rue, C. Vanderburgh, D. McKnight, and N. Fierer. 2020. Effects of Spatial Variability and Relic DNA Removal on the Detection of Temporal Dynamics in Soil Microbial Communities. mBio 11:e02776-19.. Carter, D. R., R. A. Slesak, T. B. Harrington, and A. W. D'Amato. 2019a. Comparative effects of soil resource availability on physiology and growth of Scotch broom (Cytisus scoparius) and Douglas-fir (Pseudotsuga menziesii) seedlings. Forest Ecology and Management 453,[117580]. Carter, D. R., R. A. Slesak, T. B. Harrington, and A. W. D’Amato. 2019b. Effects of irrigation and phosphorus fertilization on physiology, growth, and nitrogen-accumulation of Scotch broom (Cytisus scoparius). Plant Physiology Reports 24:410-421. Carter, D. R., R. A. Slesak, T. B. Harrington, D. H. Peter, and A. W. D’Amato. 2019c. Scotch broom (Cytisus scoparius) modifies microenvironment to promote nonnative plant communities. Biological Invasions 21:1055-1073. Cavagnaro, T., F. Smith, S. Smith, and I. Jakobsen. 2005. Functional diversity in arbuscular mycorrhizas: exploitation of soil patches with different phosphate enrichment differs among fungal species. Plant, Cell & Environment 28:642-650. Ceradini, J. P., and A. D. Chalfoun. 2017. When perception reflects reality: Non‐native grass invasion alters small mammal risk landscapes and survival. Ecology and Evolution 7:1823- 1835. Chen, Q., W. W. Wu, S. S. Qi, H. Cheng, Q. Li, Q. Ran, Z. C. Dai, D. L. Du, S. Egan, and T. Thomas. 2019a. Arbuscular mycorrhizal fungi improve the growth and disease resistance of the invasive plant Wedelia trilobata. Journal of Applied Microbiology. DOI: 10.1111/jam.14415 Chen, Z. G., K. S. Bishop, H. Tanambell, P. Buchanan, C. Smith, and S. Y. Quek. 2019b. Characterization of the bioactivities of an ethanol extract and some of its constituents from the New Zealand native mushroom Hericium novae-zealandiae. Food & Function 10:6633- 6643. Christensen, B., P. Cashmore, S. Crump, and J. Hobbs. 2019. Fire disturbance favours exotic species at Kaituna Wetland, Bay of Plenty. New Zealand Journal of Ecology 43:1-7. Chytrý, M., P. Pyšek, J. Wild, J. Pino, L. C. Maskell, and M. Vilà. 2009. European map of alien plant invasions based on the quantitative assessment across habitats. Diversity and Distributions 15:98-107. Collins, C. D., and B. L. Foster. 2009. Community‐level consequences of mycorrhizae depend on phosphorus availability. Ecology 90:2567-2576. Collins, C. G., T. F. Bohner, and J. M. Diez. 2019. Plant-soil feedbacks and facilitation influence the demography of herbaceous alpine species in response to woody plant range expansion. Frontiers in Ecology and Evolution 7:417. Collins, C. G., C. J. Carey, E. L. Aronson, C. W. Kopp, and J. M. Diez. 2016. Direct and indirect effects of native range expansion on soil microbial community structure and function. Journal of Ecology 104:1271-1283. Crawford, K. M., J. T. Bauer, L. S. Comita, M. B. Eppinga, D. J. Johnson, S. A. Mangan, S. A. Queenborough, A. E. Strand, K. N. Suding, and J. Umbanhowar. 2019. When and where plant‐soil feedback may promote plant coexistence: a meta‐analysis. Ecology Letters 22:1274-1284. Crawley, M. 2015. Statistics. An Introduction using R. 2nd rd. Wiley. Crawley, M. J. 2002. Statistical computingan introduction to data analysis using S-Plus.Wiley.

114

Crossay, T., Y. Cavaloc, C. Majorel, D. Redecker, V. Medevielle, and H. Amir. 2020. Combinations of different arbuscular mycorrhizal fungi improve fitness and metal tolerance of sorghum in ultramafic soil. Rhizosphere 14:100204. Crossay, T., C. Majorel, D. Redecker, S. Gensous, V. Medevielle, G. Durrieu, Y. Cavaloc, and H. Amir. 2019. Is a mixture of arbuscular mycorrhizal fungi better for plant growth than single-species inoculants? Mycorrhiza 29:325-339. Crowley, D., Y. Wang, C. Reid, and P. Szaniszlo. 1991. Mechanisms of iron acquisition from siderophores by microorganisms and plants. Pp. 213-232 In Iron Nutrition and Interactions in Plants. Springer. D'Antonio, C., and S. L. Flory. 2017. Long‐term dynamics and impacts of plant invasions. Journal of Ecology 105:1459-1461. Daniels, M. K., and E. R. Larson. 2020. Effects of forest windstorm disturbance on invasive plants in protected areas of southern Illinois, USA. Journal of Ecology 108:199-211. Danner, B. T., and A. K. Knapp. 2003. Abiotic constraints on the establishment of Quercus seedlings in grassland. Global Change Biology 9:266-275. Daryanto, S., P.-A. Jacinthe, B. Fu, W. Zhao, and L. Wang. 2019. Valuing the ecosystem services of cover crops: barriers and pathways forward. Agriculture, Ecosystems & Environment 270:76-78. Davis, E. J. 2018. A Widespread Nitrogen-fixing Invader Experiences Negative Soil Feedbacks Despite Increased Root Nodulation and Mycorrhizal Colonization.Undergraduate Thesis. University of California, Santa Cruz. https://guides.library.ucsc.edu/etds Davis, M., I. A. Dickie, T. Paul, and F. Carswell. 2013. Is kanuka and manuka establishment in grassland constrained by mycorrhizal abundance? New Zealand Journal of Ecology 37(2):172-177. Davison, J., M. Moora, M. Öpik, A. Adholeya, L. Ainsaar, A. Bâ, S. Burla, A. Diedhiou, I. Hiiesalu, and T. Jairus. 2015. Global assessment of arbuscular mycorrhizal fungus diversity reveals very low endemism. Science 349:970-973. Davison, J., M. Öpik, T. J. Daniell, M. Moora, and M. Zobel. 2011. Arbuscular mycorrhizal fungal communities in plant roots are not random assemblages. FEMS Microbiology Ecology 78:103-115. de Lange, P., D. Blanchon, A. Knight, J. Elix, R. Lucking, K. Frogley, A. Harris, J. Cooper, and J. R. Rolfe. 2018. Conservation status of New Zealand indigenous lichens and lichenicolous fungi. Wellington: Department of Conservation. Deiner, K., H. M. Bik, E. Mächler, M. Seymour, A. Lacoursière‐Roussel, F. Altermatt, S. Creer, I. Bista, D. M. Lodge, and N. De Vere. 2017. Environmental DNA metabarcoding: Transforming how we survey animal and plant communities. Molecular Ecology 26:5872- 5895. Detheridge, A. P., G. Brand, R. Fychan, F. V. Crotty, R. Sanderson, G. W. Griffith, and C. L. Marley. 2016. The legacy effect of cover crops on soil fungal populations in a cereal rotation. Agriculture, Ecosystems & Environment 228:49-61. Dewar, A. M., J. M. Facelli, P. Marschner, F. A. Smith, and F. D. Panetta. 2006. Gorse and broom in the Adelaide Hills: effect of invasive species on soil microbial biomass and nutrients. Pp. 203-206 In Proceedings of the Fifteenth Australian Weeds Conference; Weed Management Society of South Australia Inc. Adelaide, South Australia. Dickie, I., and P. Reich. 2005. Ectomycorrhizal fungal communities at forest edges. Journal of Ecology 93:244-255. Dickie, I., S. Richardson, and S. Wiser. 2009. Ectomycorrhizal fungal communities and soil chemistry in harvested and unharvested temperate Nothofagus rainforests. Canadian Journal of Forest Research 39:1069-1079. Dickie, I. A. 2007. Host preference, niches and fungal diversity. New Phytologist 174:230-233. Dickie, I. A. 2010. Insidious effects of sequencing errors on perceived diversity in molecular surveys. New Phytologist 188:916-918.

115

Dickie, I. A., S. Boyer, H. L. Buckley, R. P. Duncan, P. P. Gardner, I. D. Hogg, R. J. Holdaway, G. Lear, A. Makiola, and S. E. Morales. 2018. Towards robust and repeatable sampling methods in eDNA‐based studies. Molecular Ecology Resources 18:940-952. Dickie, I. A., J. L. Bufford, R. C. Cobb, M. L. Desprez‐Loustau, G. Grelet, P. E. Hulme, J. Klironomos, A. Makiola, M. A. Nuñez, and A. Pringle. 2017a. The emerging science of linked plant–fungal invasions. New Phytologist 215:1314-1332. Dickie, I. A., J. A. Cooper, J. L. Bufford, P. E. Hulme, and S. T. Bates. 2017b. Loss of functional diversity and network modularity in introduced plant–fungal symbioses. AoB Plants 9(1): plw084. Dickie, I. A., M. Davis, and F. E. Carswell. 2012. Quantification of mycorrhizal limitation in beech spread. New Zealand Journal of Ecology 36(2):210-215. Dickie, I. A., N. Koele, J. D. Blum, J. D. Gleason, and M. S. McGlone. 2014. Mycorrhizas in changing ecosystems. Botany 92:149-160. Dickie, I. A., and M. G. St John. 2016. Second-generation molecular understanding of mycorrhizas in soil ecosystems. In Molecular mycorrhizal symbiosis (edited by Francis Martin). Hoboken, New Jersey, USA, Wiley-Blackwell. Pp. 473-493. Dickie, I. A., A. Wakelin, and S. J. Richardson. 2020. Rare species of wood inhabiting fungi are not local. Ecological Applications 0(0):e02140. Dickie, I. A., A. M. Wakelin, L. B. Martínez-García, S. J. Richardson, A. Makiola, and J. M. Tylianakis. 2019. Oomycetes along a 120,000 year temperate rainforest ecosystem development chronosequence. Fungal Ecology 39:192-200. Dickie, I. A., B. Xu, and R. T. Koide. 2002. Vertical niche differentiation of ectomycorrhizal hyphae in soil as shown by T‐RFLP analysis. New Phytologist 156:527-535. Dierks, J., K. Denef, L. T. van Diepen, and M.-A. de Graaff. 2019. Cheatgrass-associated AMF community negatively affects sagebrush root production but not C transfer to the soil. Plant and Soil 436:381-396. Diez, J. M., I. Dickie, G. Edwards, P. E. Hulme, J. J. Sullivan, and R. P. Duncan. 2010. Negative soil feedbacks accumulate over time for non‐native plant species. Ecology Letters 13:803- 809. Dopheide, A., D. Xie, T. R. Buckley, A. J. Drummond, and R. D. Newcomb. 2019. Impacts of DNA extraction and PCR on DNA metabarcoding estimates of soil biodiversity. Methods in Ecology and Evolution 10:120-133. Dostálek, T., Z. Münzbergová, A. Kladivová, and M. Macel. 2016. Plant–soil feedback in native vs. invasive populations of a range expanding plant. Plant and Soil 399:209-220. Drummond, A., R. Newcomb, T. Buckley, D. Xie, A. Dopheide, B. Potter, J. Heled, H. Ross, L. Tooman, and S. Grosser. 2015. Evaluating a multigene environmental DNA approach for biodiversity assessment. GigaScience 4(1), s13742-015. Dumalisile, L., and M. J. Somers. 2017. The effects of an invasive alien plant (Chromolaena odorata) on large African mammals. Nature Conservation Research 2:102-108. During, C. 1972. Fertilisers and Soils. Government Printer, Wellington, New Zealand. Edgar, R. C. 2010. Search and clustering orders of magnitude faster than BLAST. Bioinformatics 26:2460-2461. Edwards, I. P., D. R. Zak, H. Kellner, S. D. Eisenlord, and K. S. Pregitzer. 2011. Simulated atmospheric N deposition alters fungal community composition and suppresses ligninolytic gene expression in a northern hardwood forest. PLoS One 6(6):e20421. Edwards, J., C. Santos-Medellín, B. Nguyen, J. Kilmer, Z. Liechty, E. Veliz, J. Ni, G. Phillips, and V. Sundaresan. 2019. Soil domestication by rice cultivation results in plant-soil feedback through shifts in soil microbiota. Genome Biology 20:1-14. Ehlers, G. C., J. R. Caradus, and S. V. Fowler. 2020. The regulatory process and costs to seek approval for the development and release of new biological control agents in New Zealand. BioControl 65:1-12. Ehrenfeld, J. G. 2003. Effects of exotic plant invasions on soil nutrient cycling processes. Ecosystems 6:503-523.

116

Ehrenfeld, J. G. 2010. Ecosystem consequences of biological invasions. Annual Review of Ecology, Evolution, and Systematics 41:59-80. Eisenhauer, N., H. Beßler, C. Engels, G. Gleixner, M. Habekost, A. Milcu, S. Partsch, A. C. Sabais, C. Scherber, and S. Steinbeiss. 2010. Plant diversity effects on soil microorganisms support the singular hypothesis. Ecology 91:485-496. Ellingsøe, P., and K. Johnsen. 2002. Influence of soil sample sizes on the assessment of bacterial community structure. Soil Biology and Biochemistry 34:1701-1707. Emer, C., J. Memmott, I. P. Vaughan, D. Montoya, and J. M. Tylianakis. 2016. Species roles in plant–pollinator communities are conserved across native and alien ranges. Diversity and Distributions 22:841-852. Emerson, W. 1995. Water-retention, organic-C and soil texture. Soil Research 33:241-251. Essl, F., W. Dawson, H. Kreft, J. Pergl, P. Pyšek, M. Van Kleunen, P. Weigelt, T. Mang, S. Dullinger, and B. Lenzner. 2019. Drivers of the relative richness of naturalized and invasive plant species on Earth. AoB Plants 11:plz051. Ettema, C. H., and D. A. Wardle. 2002. Spatial soil ecology. Trends in Ecology & Evolution 17:177-183. Evgrafova, A., T. R. de la Haye, I. Haase, O. Shibistova, G. Guggenberger, N. Tananaev, L. Sauheitl, and S. Spielvogel. 2018. Small-scale spatial patterns of soil organic carbon and nitrogen stocks in permafrost-affected soils of northern Siberia. Geoderma 329:91-107. Fagerström, T., and U. Lohm. 1977. Growth in Scots pine (Pinus silvestris L.). Oecologia 26:305- 315. Fahey, C., C. Angelini, and S. L. Flory. 2018. Grass invasion and drought interact to alter the diversity and structure of native plant communities. Ecology 99:2692-2702. Farr, D. F., G. F. Bills, G. P. Chamuris, and A. Y. Rossman. 1989. Fungi on plants and plant products in the United States. APS Press, USA. Fernandez, R. D., N. Bulacio, A. Álvarez, H. Pajot, and R. Aragón. 2017. Fungal decomposers of leaf litter from an invaded and native mountain forest of NW Argentina. Antonie van Leeuwenhoek 110:1207-1218. Fierer, N., and R. B. Jackson. 2006. The diversity and biogeography of soil bacterial communities. Proceedings of the National Academy of Sciences 103:626-631. Figueiral, I., and A. M. Bettencourt. 2004. Middle/Late Bronze Age plant communities and their exploitation in the Cavado Basin (NW Portugal) as shown by charcoal analysis: the significance and co-occurrence of Quercus (deciduous)–Fabaceae. Vegetation History and Archaeobotany 13:219-232. Flynn, J. M., E. A. Brown, F. J. Chain, H. J. MacIsaac, and M. E. Cristescu. 2015. Toward accurate molecular identification of species in complex environmental samples: testing the performance of sequence filtering and clustering methods. Ecology and Evolution 5:2252- 2266. Fogarty, G., and J. M. Facelli. 1999. Growth and competition of Cytisus scoparius, an invasive shrub, and Australian native shrubs. Plant Ecology 144:27-35. Foster, K. R., and T. Bell. 2012. Competition, not cooperation, dominates interactions among culturable microbial species. Current biology 22:1845-1850. Fuller, R., P. Buchanan, and M. Roberts. 2004. Maori knowledge of fungi/Matauranga o nga harore. The Fungi of New Zealand/Ng ā Harore o Aotearoa 1:81-118. Gaggini, L., H.-P. Rusterholz, and B. Baur. 2018. The invasive plant Impatiens glandulifera affects soil fungal diversity and the bacterial community in forests. Applied Soil Ecology 124:335- 343. Gao, C., N. N. Shi, Y. X. Liu, K. G. Peay, Y. Zheng, Q. Ding, X. C. Mi, K. P. Ma, T. Wubet, and F. Buscot. 2013. Host plant genus‐level diversity is the best predictor of ectomycorrhizal fungal diversity in a Chinese subtropical forest. Molecular Ecology 22:3403-3414. García‐Guzmán, G., and M. Heil. 2014. Life histories of hosts and pathogens predict patterns in tropical fungal plant diseases. New Phytologist 201:1106-1120. Gaston, K., and T. Blackburn. 2008. Pattern and process in macroecology. John Wiley & Sons.

117

Gaston, K. J. 2000. Global patterns in biodiversity. Nature 405:220-227. Ghanizadeh, H., and K. C. Harrington. 2019. Weed management in New Zealand pastures. Agronomy 9(8):448. Gibbons, S. M., Y. Lekberg, D. L. Mummey, N. Sangwan, P. W. Ramsey, and J. A. Gilbert. 2017. Invasive plants rapidly reshape soil properties in a grassland ecosystem. MSystems 2:e00178-16. Gilbert, J. A., J. K. Jansson, and R. Knight. 2014. The Earth Microbiome project: successes and aspirations. BMC Biology 12:69. Gilliam, F. S., B. M. Yurish, and M. B. Adams. 2001. Temporal and spatial variation of nitrogen transformations in nitrogen-saturated soils of a central Appalachian hardwood forest. Canadian Journal of Forest Research 31:1768-1785. Gornish, E. S., N. Fierer, and A. Barberán. 2016. Associations between an invasive plant (Taeniatherum caput-medusae, Medusahead) and soil microbial communities. PLoS One 11(9):e0163930. Gossner, M. M., T. M. Lewinsohn, T. Kahl, F. Grassein, S. Boch, D. Prati, K. Birkhofer, S. C. Renner, J. Sikorski, and T. Wubet. 2016. Land-use intensification causes multitrophic homogenization of grassland communities. Nature 540:266-269. Green, J. L., A. J. Holmes, M. Westoby, I. Oliver, D. Briscoe, M. Dangerfield, M. Gillings, and A. J. Beattie. 2004. Spatial scaling of microbial eukaryote diversity. Nature 432:747-750. Grotkopp, E., and M. Rejmánek. 2007. High seedling relative growth rate and specific leaf area are traits of invasive species: phylogenetically independent contrasts of woody angiosperms. American Journal of Botany 94:526-532. Grove, S., K. A. Haubensak, C. Gehring, and I. M. Parker. 2017. Mycorrhizae, invasions, and the temporal dynamics of mutualism disruption. Journal of Ecology 105:1496-1508. Grove, S., K. A. Haubensak, and I. M. Parker. 2012. Direct and indirect effects of allelopathy in the soil legacy of an exotic plant invasion. Plant Ecology 213:1869-1882. Grove, S., I. M. Parker, and K. A. Haubensak. 2015. Persistence of a soil legacy following removal of a nitrogen-fixing invader. Biological Invasions 17:2621-2631. Grzędzicka, E., and J. Reif. 2020. Impacts of an invasive plant on bird communities differ along a habitat gradient. Global Ecology and Conservation 23:e01150. Guerin-Laguette, A., R. Butler, and Y. Wang. 2020. Advances in the Cultivation of Lactarius deliciosus (Saffron Milk Cap) in New Zealand. Pp. 141-161 Mushrooms, Humans and Nature in a Changing World. Springer. Gui, H., K. Hyde, J. Xu, and P. Mortimer. 2017. Arbuscular mycorrhiza enhance the rate of litter decomposition while inhibiting soil microbial community development. Scientific Reports 7:42184. Gupta, S., and W. Larson. 1979. Estimating soil water retention characteristics from particle size distribution, organic matter percent, and bulk density. Water Resources Research 15:1633- 1635. Han, W., J. Fang, D. Guo, and Y. Zhang. 2005. Leaf nitrogen and phosphorus stoichiometry across 753 terrestrial plant species in China. New Phytologist 168:377-385. Hantsch, L., U. Braun, J. Haase, O. Purschke, M. Scherer-Lorenzen, and H. Bruelheide. 2014. No plant functional diversity effects on foliar fungal pathogens in experimental tree communities. Fungal Diversity 66:139-151. Hantsch, L., U. Braun, M. Scherer-Lorenzen, and H. Bruelheide. 2013. Species richness and species identity effects on occurrence of foliar fungal pathogens in a tree diversity experiment. Ecosphere 4:1-12. Harrington, T. 2011. Quantifying competitive ability of perennial grasses to inhibit Scotch broom. Research Paper PNW-RP-587. US Department of Agriculture, Forest Service, Pacific Northwest Research Station. Harrington, T. B., D. H. Peter, and R. A. Slesak. 2018. Logging debris and herbicide treatments improve growing conditions for planted Douglas-fir on a droughty forest site invaded by Scotch broom. Forest Ecology and Management 417:31-39.

118

Hartnett, D. C., and G. W. Wilson. 1999. Mycorrhizae influence plant community structure and diversity in tallgrass prairie. Ecology 80:1187-1195. Haubensak, K. A., S. Grove, J. Foster, and I. M. Parker. 2020. Chemical and mechanical control of the invasive shrub Cytisus scoparius in forest clearings in western Washington, USA. Invasive Plant Science and Management 13:30-36. Haubensak, K. A., and I. M. Parker. 2004. Soil changes accompanying invasion of the exotic shrub Cytisus scoparius in glacial outwash prairies of western Washington, USA. Plant Ecology 175:71-79. Hawkes, C. V. 2007. Are invaders moving targets? The generality and persistence of advantages in size, reproduction, and enemy release in invasive plant species with time since introduction. The American Naturalist 170:832-843. Hawkes, C. V., J. Belnap, C. D’Antonio, and M. K. Firestone. 2006. Arbuscular mycorrhizal assemblages in native plant roots change in the presence of invasive exotic grasses. Plant and Soil 281:369-380. Hawkes, C. V., S. N. Kivlin, J. Du, and V. T. Eviner. 2013. The temporal development and additivity of plant-soil feedback in perennial grasses. Plant and Soil 369:141-150. Hawksworth, D. 2012. Global species numbers of fungi: are tropical studies and molecular approaches contributing to a more robust estimate? Biodiversity and Conservation 21:2425-2433. Hayward, J., T. R. Horton, A. Pauchard, and M. A. Nuñez. 2015. A single ectomycorrhizal fungal species can enable a Pinus invasion. Ecology 96:1438-1444. Heenan, P. B. 1998. Phylogenetic analysis of the Carmichaelia complex, Clianthus, and Swainsona (Fabaceae), from Australia and New Zealand. New Zealand Journal of Botany 36:21-40. Heenan, P. B., P. J., de Lange, and A. D. Wilton. 2001. Sophora (Fabaceae) in New Zealand: taxonomy, distribution, and biogeography New Zealand Journal of Botany 39:17-34. Heinen, R., S. E. Hannula, J. R. De Long, M. Huberty, R. Jongen, A. Kielak, K. Steinauer, F. Zhu, and T. M. Bezemer. 2020. Plant community composition steers grassland vegetation via soil legacy effects. Ecology Letters 23:973-982. Hejda, M., P. Pyšek, and V. Jarošík. 2009. Impact of invasive plants on the species richness, diversity and composition of invaded communities. Journal of Ecology 97:393-403. Helgason, T., J. Merryweather, J. Denison, P. Wilson, J. P. W. Young, and A. Fitter. 2002. Selectivity and functional diversity in arbuscular mycorrhizas of co‐occurring fungi and plants from a temperate deciduous woodland. Journal of Ecology 90:371-384. Helgerson, O., J. Gordon, and D. Perry. 1984. N2 fixation by red alder (Alnus rubra) and scotch broom (Cytisus scoparius) planted under precommerically thinned Douglas-fir (Pseudotsuga menziesii). Plant and Soil 78:221-233. Hermans, S. M., H. L. Buckley, and G. Lear. 2018. Optimal extraction methods for the simultaneous analysis of DNA from diverse organisms and sample types. Molecular Ecology Resources 18:557-569. Hierro, J. L., J. L. Maron, and R. M. Callaway. 2005. A biogeographical approach to plant invasions: the importance of studying exotics in their introduced and native range. Journal of Ecology 93:5-15. Higo, M., R. Sato, A. Serizawa, Y. Takahashi, K. Gunji, Y. Tatewaki, and K. Isobe. 2018. Can phosphorus application and cover cropping alter arbuscular mycorrhizal fungal communities and soybean performance after a five-year phosphorus-unfertilized crop rotational system? PeerJ 6:e4606. Hiiesalu, I., M. Pärtel, J. Davison, P. Gerhold, M. Metsis, M. Moora, M. Öpik, M. Vasar, M. Zobel, and S. D. Wilson. 2014. Species richness of arbuscular mycorrhizal fungi: associations with grassland plant richness and biomass. New Phytologist 203:233-244. Hill, M. P., V. C. Moran, J. H. Hoffmann, S. Neser, H. G. Zimmermann, D. O. Simelane, H. Klein, C. Zachariades, A. R. Wood, and M. J. Byrne. 2020. More than a century of biological control against invasive alien plants in South Africa: a synoptic view of what has been accomplished. Pp. 553-572. Biological Invasions in South Africa. Springer.

119

Hiraoka, S., C.-c. Yang, and W. Iwasaki. 2016. Metagenomics and bioinformatics in microbial ecology: current status and beyond. Microbes and Environments 0(0):ME16024. Hodge, A., C. D. Campbell, and A. H. Fitter. 2001. An arbuscular mycorrhizal fungus accelerates decomposition and acquires nitrogen directly from organic material. Nature 413:297-299. Hoeksema, J. D., J. D. Bever, S. Chakraborty, V. B. Chaudhary, M. Gardes, C. A. Gehring, M. M. Hart, E. A. Housworth, W. Kaonongbua, and J. N. Klironomos. 2018. Evolutionary history of plant hosts and fungal symbionts predicts the strength of mycorrhizal mutualism. Communications Biology 1:1-10. Holdaway, R. J., J. R. Wood, I. A. Dickie, K. H. Orwin, P. J. Bellingham, S. J. Richardson, P. O. B. Lyver, P. Timoti, and T. R. Buckley. 2017. Using DNA metabarcoding to assess New Zealand’s terrestrial biodiversity. New Zealand Journal of Ecology 41:251-262. Holm, L. G., L. Holm, E. Holm, J. V. Pancho, and J. P. Herberger. 1997. World weeds: natural histories and distribution. John Wiley & Sons. Horton, T. R., and T. D. Bruns. 2001. The molecular revolution in ectomycorrhizal ecology: peeking into the black‐box. Molecular Ecology 10:1855-1871. Horwath, W. R. 2017. The role of the soil microbial biomass in cycling nutrients. Microbial Biomass: A Paradigm Shift in Terrestrial Biogeochemistry. World Scientific:41-66. Howell, C. J., and J. A. Terry. 2016. The creation of a New Zealand weed atlas. Department of Conservation. Hu, F. S., B. P. Finney, and L. B. Brubaker. 2001. Effects of Holocene Alnus expansion on aquatic productivity, nitrogen cycling, and soil development in southwestern Alaska. Ecosystems 4:358-368. Hulme, P. E. 2020. Plant invasions in New Zealand: global lessons in prevention, eradication and control. Biological Invasions 22:1539-1562. Hurst, J., and R. Allen. 1993. A permanent plot method for monitoring indigenous forests: field protocols. Landcare Research New Zealand. Ihrmark, K., I. Bödeker, K. Cruz-Martinez, H. Friberg, A. Kubartova, J. Schenck, Y. Strid, J. Stenlid, M. Brandström-Durling, and K. E. Clemmensen. 2012. New primers to amplify the fungal ITS2 region–evaluation by 454-sequencing of artificial and natural communities. FEMS Microbiology Ecology 82:666-677. Ingraffia, R., G. Amato, A. S. Frenda, and D. Giambalvo. 2019. Impacts of arbuscular mycorrhizal fungi on nutrient uptake, N2 fixation, N transfer, and growth in a wheat/faba bean intercropping system. PLoS One 14(3): e0213672. Innes, L., P. J. Hobbs, and R. D. Bardgett. 2004. The impacts of individual plant species on rhizosphere microbial communities in soils of different fertility. Biology and Fertility of Soils 40:7-13. Jansa, J., F. A. Smith, and S. E. Smith. 2008. Are there benefits of simultaneous root colonization by different arbuscular mycorrhizal fungi? New Phytologist 177:779-789. Jarvis, P., S. Fowler, Q. Paynter, and P. Syrett. 2006. Predicting the economic benefits and costs of introducing new biological control agents for Scotch broom Cytisus scoparius into New Zealand. Biological Control 39:135-146. Jelbert, K., D. Buss, J. McDonald, S. Townley, M. Franco, I. Stott, O. Jones, R. Salguero-Gómez, Y. Buckley, and T. Knight. 2019. Demographic amplification is a predictor of invasiveness among plants. Nature Communications 10:1-6. Jesse, W. A., J. Molleman, O. Franken, M. Lammers, M. P. Berg, J. E. Behm, M. R. Helmus, and J. Ellers. 2020. Disentangling the effects of plant species invasion and urban development on arthropod community composition. Global Change Biology 26:3294- 3306. Jiang, J., K. A. Alderisio, A. Singh, and L. Xiao. 2005. Development of procedures for direct extraction of Cryptosporidium DNA from water concentrates and for relief of PCR inhibitors. Applied and Environmental Microbiology 71:1135-1141. Johnston, P. R., Parkes, S.L., and Broadhurst, P.G. 1995. Fungi associated with gorse and broom in New Zealand. Australasian Plant Pathology 24: 157–167.

120

Jones, M. D., and S. E. Smith. 2004. Exploring functional definitions of mycorrhizas: are mycorrhizas always mutualisms? Canadian Journal of Botany 82:1089-1109. Jordan, N. R., L. Aldrich-Wolfe, S. C. Huerd, D. L. Larson, and G. Muehlbauer. 2012. Soil– occupancy effects of invasive and native grassland plant species on composition and diversity of mycorrhizal associations. Invasive Plant Science and Management 5:494-505. Kadowaki, K., S. Yamamoto, H. Sato, A. S. Tanabe, A. Hidaka, and H. Toju. 2018. Mycorrhizal fungi mediate the direction and strength of plant–soil feedbacks differently between arbuscular mycorrhizal and ectomycorrhizal communities. Communications Biology 1:1- 11. Kamiya, T., K. O'Dwyer, S. Nakagawa, and R. Poulin. 2014. What determines species richness of parasitic organisms? A meta‐analysis across animal, plant and fungal hosts. Biological Reviews 89:123-134. Kang, S., and A. L. Mills. 2006. The effect of sample size in studies of soil microbial community structure. Journal of microbiological methods 66:242-250. Kardol, P., G. B. De Deyn, E. Laliberté, P. Mariotte, and C. V. Hawkes. 2013. Biotic plant–soil feedbacks across temporal scales. Journal of Ecology 101:309-315. Kardol, P., J. R. De Long, and M. K. Sundqvist. 2012. Crossing the threshold: the power of multi- level experiments in identifying global change responses. New Phytologist 196:323-326. Keet, J.-H., A. G. Ellis, C. Hui, and J. J. Le Roux. 2017. Legume–rhizobium symbiotic promiscuity and effectiveness do not affect plant invasiveness. Annals of Botany 119:1319- 1331. Kempel, A., A. Rindisbacher, M. Fischer, and E. Allan. 2018. Plant soil feedback strength in relation to large‐scale plant rarity and phylogenetic relatedness. Ecology 99:597-606. Kiers, E. T., M. Duhamel, Y. Beesetty, J. A. Mensah, O. Franken, E. Verbruggen, C. R. Fellbaum, G. A. Kowalchuk, M. M. Hart, and A. Bago. 2011. Reciprocal rewards stabilize cooperation in the mycorrhizal symbiosis. Science 333:880-882. Kivlin, S. N., R. Bedoya, and C. V. Hawkes. 2018. Heterogeneity in arbuscular mycorrhizal fungal communities may contribute to inconsistent plant-soil feedback in a Neotropical forest. Plant and Soil 432:29-44. Kivlin, S. N., C. V. Hawkes, and K. K. Treseder. 2011. Global diversity and distribution of arbuscular mycorrhizal fungi. Soil Biology and Biochemistry 43:2294-2303. Klaubauf, S., E. Inselsbacher, S. Zechmeister-Boltenstern, W. Wanek, R. Gottsberger, J. Strauss, and M. Gorfer. 2010. Molecular diversity of fungal communities in agricultural soils from Lower Austria. Fungal Diversity 44:65-75. Klingmüller, W., A. Dally, C. Fentner, and M. Steinlein. 1990. Plasmid transfer between soil bacteria. Pages 133-151 Bacterial Genetics in Natural Environments. Springer. Klironomos, J. N. 2002. Feedback with soil biota contributes to plant rarity and invasiveness in communities. Nature 417:67. Klironomos, J. N. 2003. Variation in plant response to native and exotic arbuscular mycorrhizal fungi. Ecology 84:2292-2301. Klironomos, J. N., J. McCune, M. Hart, and J. Neville. 2000. The influence of arbuscular mycorrhizae on the relationship between plant diversity and productivity. Ecology Letters 3:137-141. Klock, M. M., Barrett, L. G., Thrall, P. H. and Harms, K. E. 2015. Host promiscuity in symbiont associations can influence exotic legume establishment and colonization of novel ranges. Diversity and Distributions. 21: 1193–1203. Knight, S. J., O. Karon, and M. R. Goddard. 2020. Small scale fungal community differentiation in a vineyard system. Food Microbiology 87:103358. Koch, A. M., P. M. Antunes, H. Maherali, M. M. Hart, and J. N. Klironomos. 2017. Evolutionary asymmetry in the arbuscular mycorrhizal symbiosis: conservatism in fungal morphology does not predict host plant growth. New Phytologist 214:1330-1337. Koerselman, W., and A. F. Meuleman. 1996. The vegetation N: P ratio: a new tool to detect the nature of nutrient limitation. Journal of Applied Ecology 33(6):1441-1450.

121

Kohout, P., R. Sudová, M. Janoušková, M. Čtvrtlíková, M. Hejda, H. Pánková, R. Slavíková, K. Štajerová, M. Vosátka, and Z. Sýkorová. 2014. Comparison of commonly used primer sets for evaluating arbuscular mycorrhizal fungal communities: Is there a universal solution? Soil Biology and Biochemistry 68:482-493. Kowalchuk, G., W. G. Hol, and J. Van Veen. 2006. Rhizosphere fungal communities are influenced by Senecio jacobaea pyrrolizidine alkaloid content and composition. Soil Biology and Biochemistry 38:2852-2859. Koziol, A., M. Stat, T. Simpson, S. Jarman, J. D. DiBattista, E. S. Harvey, M. Marnane, J. McDonald, and M. Bunce. 2019. Environmental DNA metabarcoding studies are critically affected by substrate selection. Molecular Ecology Resources 19:366-376. Kreader, C. A. 1996. Relief of amplification inhibition in PCR with bovine serum albumin or T4 gene 32 protein. Applied and Environmental Microbiology 62:1102-1106. Kuchár, M., T. R. Glare, J. G. Hampton, I. A. Dickie, and M. C. Christey. 2019. Virulence of the plant-associated endophytic fungus Lecanicillium muscarium to diamondback moth larvae. New Zealand Plant Protection 72:253-259. Kulmatiski, A., and K. H. Beard. 2011. Long-term plant growth legacies overwhelm short-term plant growth effects on soil microbial community structure. Soil Biology and Biochemistry 43:823-830. Kulmatiski, A., K. H. Beard, J. R. Stevens, and S. M. Cobbold. 2008. Plant–soil feedbacks: a meta‐ analytical review. Ecology Letters 11:980-992. Lang, C., J. Seven, and A. Polle. 2011. Host preferences and differential contributions of deciduous tree species shape mycorrhizal species richness in a mixed Central European forest. Mycorrhiza 21:297-308. Lang, M., H. M. Hanslin, J. Kollmann, and T. Wagner. 2017. Suppression of an invasive legume by a native grass—High impact of priority effects. Basic and Applied Ecology 22:20-27. Larsen, B. B., E. C. Miller, M. K. Rhodes, and J. J. Wiens. 2017. Inordinate fondness multiplied and redistributed: the number of species on Earth and the new pie of life. The Quarterly Review of Biology 92:229-265. Latz, E., N. Eisenhauer, B. C. Rall, S. Scheu, and A. Jousset. 2016. Unravelling linkages between plant community composition and the pathogen-suppressive potential of soils. Scientific Reports 6:23584. Lawton, J. H., D. Bignell, B. Bolton, G. Bloemers, P. Eggleton, P. Hammond, M. Hodda, R. Holt, T. Larsen, and N. Mawdsley. 1998. Biodiversity inventories, indicator taxa and effects of habitat modification in tropical forest. Nature 391:72-76. Lear, G., I. Dickie, J. Banks, S. Boyer, H. L. Buckley, T. R. Buckley, R. Cruickshank, A. Dopheide, K. M. Handley, and S. Hermans. 2018. Methods for the extraction, storage, amplification and sequencing of DNA from environmental samples. New Zealand Journal of Ecology 42(1), pp. 10, 1A-50A. Lee, W., R. Allen, and P. Johnson. 1986. Succession and dynamics of gorse (Ulex europaeus L.) communities in the Dunedin ecological district south island, New Zealand. New Zealand Journal of Botany 24:279-292. Lefebvre, B. 2019. An opportunity to breed rice for improved benefits from the arbuscular mycorrhizal symbiosis? New Phytologist 225(4): 1404-1406. Lekberg, Y., S. M. Gibbons, S. Rosendahl, and P. W. Ramsey. 2013. Severe plant invasions can increase mycorrhizal fungal abundance and diversity. The ISME journal 7:1424-1433. Lekberg, Y., and R. Koide. 2005. Is plant performance limited by abundance of arbuscular mycorrhizal fungi? A meta‐analysis of studies published between 1988 and 2003. New Phytologist 168:189-204. Lennon, J. T. 2011. Replication, lies and lesser‐known truths regarding experimental design in environmental microbiology. Environmental Microbiology 13:1383-1386. Leray, M., and N. Knowlton. 2015. DNA barcoding and metabarcoding of standardized samples reveal patterns of marine benthic diversity. Proceedings of the National Academy of Sciences 112:2076-2081.

122

Leray, M., and N. Knowlton. 2017. Random sampling causes the low reproducibility of rare eukaryotic OTUs in Illumina COI metabarcoding. PeerJ 5:e3006. Liao, C., R. Peng, Y. Luo, X. Zhou, X. Wu, C. Fang, J. Chen, and B. Li. 2008. Altered ecosystem carbon and nitrogen cycles by plant invasion: a meta‐analysis. New Phytologist 177:706- 714. Lin, X., Y. Feng, H. Zhang, R. Chen, J. Wang, J. Zhang, and H. Chu. 2012. Long-term balanced fertilization decreases arbuscular mycorrhizal fungal diversity in an arable soil in North China revealed by 454 pyrosequencing. Environmental Science & Technology 46:5764- 5771. Lindahl, B. D., R. H. Nilsson, L. Tedersoo, K. Abarenkov, T. Carlsen, R. Kjøller, U. Kõljalg, T. Pennanen, S. Rosendahl, and J. Stenlid. 2013. Fungal community analysis by high‐ throughput sequencing of amplified markers–a user's guide. New Phytologist 199:288-299. Lishawa, S. C., B. A. Lawrence, D. A. Albert, D. J. Larkin, and N. C. Tuchman. 2019. Invasive species removal increases species and phylogenetic diversity of wetland plant communities. Ecology and Evolution 9:6231-6244. Liu, J., Z. Yang, P. Dang, H. Zhu, Y. Gao, V. N. Ha, and Z. Zhao. 2018. Response of soil microbial community dynamics to Robinia pseudoacacia L. afforestation in the loess plateau: a chronosequence approach. Plant and Soil 423:327-338. Lu, X., M. He, J. Ding, and E. Siemann. 2018. Latitudinal variation in soil biota: testing the biotic interaction hypothesis with an invasive plant and a native congener. The ISME journal 12:2811-2822. Magda, D., B. Gleizes, and M. Jarry. 2013. Maternal effect on seed survival and emergence in Cytisus scoparius: an experimental approach. Ecological Research 28:927-934. Makiola, A., I. A. Dickie, R. J. Holdaway, J. R. Wood, K. H. Orwin, and T. R. Glare. 2019a. Land use is a determinant of plant pathogen alpha‐but not beta‐diversity. Molecular Ecology 28:3786-3798. Makiola, A., I. A. Dickie, R. J. Holdaway, J. R. Wood, K. H. Orwin, C. K. Lee, and T. R. Glare. 2019b. Biases in the metabarcoding of plant pathogens using rust fungi as a model system. MicrobiologyOpen 8:e00780. Malarczyk, D., J. Panek, and M. Frąc. 2019. Alternative Molecular-Based Diagnostic Methods of Plant Pathogenic Fungi Affecting Berry Crops—A Review. Molecules 24(7):1200. Mangan, S. A., S. A. Schnitzer, E. A. Herre, K. M. Mack, M. C. Valencia, E. I. Sanchez, and J. D. Bever. 2010. Negative plant–soil feedback predicts tree-species relative abundance in a tropical forest. Nature 466:752-755. Mangelsdorff, R., M. Piepenbring, and O. Perdomo-Sánchez. 2012. Correlation of diversity of rust fungi and their host plants with disturbance and conservation of vegetation in western Panama. Biodiversity and Conservation 21:2323-2339. Mangla, S., Inderjit, and R. M. Callaway. 2008. Exotic invasive plant accumulates native soil pathogens which inhibit native plants. Journal of Ecology 96:58-67. Manoharan, L., N. P. Rosenstock, A. Williams, and K. Hedlund. 2017. Agricultural management practices influence AMF diversity and community composition with cascading effects on plant productivity. Applied Soil Ecology 115:53-59. Manter, D. K., T. L. Weir, and J. M. Vivanco. 2010. Negative effects of sample pooling on PCR- based estimates of soil microbial richness and community structure. Applied and Environmental Microbiology 76:2086-2090. Maraun, M., and S. Scheu. 1996. Changes in microbial biomass, respiration and nutrient status of beech (Fagus sylvatica) leaf litter processed by millipedes (Glomeris marginata). Oecologia 107:131-140. Mariotte, P., Z. Mehrabi, T. M. Bezemer, G. B. De Deyn, A. Kulmatiski, B. Drigo, G. C. Veen, M. G. Van der Heijden, and P. Kardol. 2018. Plant–soil feedback: bridging natural and agricultural sciences. Trends in Ecology & Evolution 33:129-142. Marler, M. J., C. A. Zabinski, and R. M. Callaway. 1999. Mycorrhizae indirectly enhance competitive effects of an invasive forb on a native bunchgrass. Ecology 80:1180-1186.

123

Maron, J. L., M. Marler, J. N. Klironomos, and C. C. Cleveland. 2011. Soil fungal pathogens and the relationship between plant diversity and productivity. Ecology Letters 14:36-41. Martínez‐García, L. B., S. J. Richardson, J. M. Tylianakis, D. A. Peltzer, and I. A. Dickie. 2015. Host identity is a dominant driver of mycorrhizal fungal community composition during ecosystem development. New Phytologist 205:1565-1576. Maul, J., and L. Drinkwater. 2010. Short-term plant species impact on microbial community structure in soils with long-term agricultural history. Plant and Soil 330:369-382. McCartney, H. A., S. J. Foster, B. A. Fraaije, and E. Ward. 2003. Molecular diagnostics for fungal plant pathogens. Pest Management Science 59:129-142. McKenzie, E., P. Johnston, and P. Buchanan. 2006. Checklist of fungi on teatree (Kunzea and Leptospermum species) in New Zealand. New Zealand Journal of Botany 44:293-335. Mendes, R., M. Kruijt, I. De Bruijn, E. Dekkers, M. van der Voort, J. H. Schneider, Y. M. Piceno, T. Z. DeSantis, G. L. Andersen, and P. A. Bakker. 2011. Deciphering the rhizosphere microbiome for disease-suppressive bacteria. Science 332:1097-1100. Mengel, K., and E. Kirkby. 1982. Principles of plant nutrition. International Potash Institute. Bern, Switzerland. Menzel, A., S. Hempel, S. Klotz, M. Moora, P. Pyšek, M. C. Rillig, M. Zobel, and I. Kühn. 2017. Mycorrhizal status helps explain invasion success of alien plant species. Ecology 98:92- 102. Míguez-Montero, M., A. Valentine, and M. Pérez-Fernández. 2020. Regulatory effect of phosphorus and nitrogen on nodulation and plant performance of leguminous shrubs. AoB Plants 12:plz047. Millar, R. B., and M. J. Anderson. 2004. Remedies for pseudoreplication. Fisheries Research 70:397-407. Miransari, M., H. Bahrami, F. Rejali, and M. Malakouti. 2009. Effects of soil compaction and arbuscular mycorrhiza on corn (Zea mays L.) nutrient uptake. Soil and Tillage Research 103:282-290. Mitchell, C. E., D. Blumenthal, V. Jarošík, E. E. Puckett, and P. Pyšek. 2010. Controls on pathogen species richness in plants’ introduced and native ranges: roles of residence time, range size and host traits. Ecology Letters 13:1525-1535. Mitchell, C. E., and A. G. Power. 2003. Release of invasive plants from fungal and viral pathogens. Nature 421:625-627. Mooney, H. A., and R. J. Hobbs. 2000. Invasive species in a changing world. Island Press. Washington, US. Morton, J. B., and D. Redecker. 2001. Two new families of Glomales, Archaeosporaceae and Paraglomaceae, with two new genera Archaeospora and Paraglomus, based on concordant molecular and morphological characters. Mycologia 93:181-195. Munkvold, L., R. Kjøller, M. Vestberg, S. Rosendahl, and I. Jakobsen. 2004. High functional diversity within species of arbuscular mycorrhizal fungi. New Phytologist 164:357-364. Nambiar, E. S., and R. Sands. 1993. Competition for water and nutrients in forests. Canadian Journal of Forest Research 23:1955-1968. Navarrete, A. A., S. M. Tsai, L. W. Mendes, K. Faust, M. de Hollander, N. A. Cassman, J. Raes, J. A. van Veen, and E. E. Kuramae. 2015. Soil microbiome responses to the short‐term effects of Amazonian deforestation. Molecular Ecology 24:2433-2448. Newsham, K., A. Fitter, and A. Watkinson. 1995. Arbuscular mycorrhiza protect an annual grass from root pathogenic fungi in the field. Journal of Ecology 83:991-1000. Nguyen, N. H., Z. Song, S. T. Bates, S. Branco, L. Tedersoo, J. Menke, J. S. Schilling, and P. G. Kennedy. 2016. FUNGuild: an open annotation tool for parsing fungal community datasets by ecological guild. Fungal Ecology 20:241-248. Nilsen, E. T., D. Karpa, H. Mooney, and C. Field. 1993. Patterns of stem photosynthesis in two invasive legumes (Spartium junceum, Cytisus scoparius) of the California coastal region. American Journal of Botany 80:1126-1136.

124

Nilsson, R. H., S. Anslan, M. Bahram, C. Wurzbacher, P. Baldrian, and L. Tedersoo. 2019a. Mycobiome diversity: high-throughput sequencing and identification of fungi. Nature Reviews Microbiology 17:95-109. Nilsson, R. H., K.-H. Larsson, A. F. S. Taylor, J. Bengtsson-Palme, T. S. Jeppesen, D. Schigel, P. Kennedy, K. Picard, F. O. Glöckner, and L. Tedersoo. 2019b. The UNITE database for molecular identification of fungi: handling dark taxa and parallel taxonomic classifications. Nucleic Acids Research 47:D259-D264. Nunes, A. L., J. M. Fill, S. J. Davies, M. Louw, A. D. Rebelo, C. J. Thorp, G. Vimercati, and J. Measey. 2019. A global meta-analysis of the ecological impacts of alien species on native amphibians. Proceedings of the Royal Society B 286:20182528. Nuñez, M. A., and I. A. Dickie. 2014. Invasive belowground mutualists of woody plants. Biological Invasions 16:645-661. Oba, H., K. Tawaray, and T. Wagatsuma. 2001. Arbuscular mycorrhizal colonization in Lupinus and related genera. Soil Science and Plant Nutrition 47:685-694. Ohman, M. D., and B. E. Lavaniegos. 2002. Comparative zooplankton sampling efficiency of a ring net and bongo net with comments on pooling of subsamples. California Cooperative Oceanic Fisheries Investigations Report 43:162-173. Oksanen, J., F. G. Blanchet, R. Kindt, P. Legendre, P. R. Minchin, R. O’hara, G. L. Simpson, P. Solymos, M. H. H. Stevens, and H. Wagner. 2013. Package ‘vegan’. Community ecology package, version 2:1-295. Olden, J. D. 2006. Biotic homogenization: a new research agenda for conservation biogeography. Journal of Biogeography 33:2027-2039. Öpik, M., M. Moora, J. Liira, and M. Zobel. 2006. Composition of root‐colonizing arbuscular mycorrhizal fungal communities in different ecosystems around the globe. Journal of Ecology 94:778-790. Öpik, M., M. Moora, M. Zobel, Ü. Saks, R. Wheatley, F. Wright, and T. Daniell. 2008. High diversity of arbuscular mycorrhizal fungi in a boreal herb‐rich coniferous forest. New Phytologist 179:867-876. Orchard, S., S. Hilton, G. D. Bending, I. A. Dickie, R. J. Standish, D. B. Gleeson, R. P. Jeffery, J. R. Powell, C. Walker, and D. Bass. 2017. Fine endophytes (Glomus tenue) are related to Mucoromycotina, not Glomeromycota. New Phytologist 213:481-486. Osborne, C. A., A. B. Zwart, L. M. Broadhurst, A. G. Young, and A. E. Richardson. 2011. The influence of sampling strategies and spatial variation on the detected soil bacterial communities under three different land-use types. FEMS Microbiology Ecology 78:70-79. Owen, S. 1998. Department of Conservation strategic plan for managing invasive weeds. Department of Conservation, Wellington, New Zealand. Packer, A., and K. Clay. 2000. Soil pathogens and spatial patterns of seedling mortality in a temperate tree. Nature 404:278. Pan, J., C. Huang, F. Peng, W. Zhang, J. Luo, S. Ma, and X. Xue. 2020. Effect of arbuscular mycorrhizal fungi (AMF) and plant growth-promoting bacteria (PGPR) inoculations on Elaeagnus Angustifolia L. in saline soil. Applied Sciences 10:945. Pardo-Muras, M., C. G. Puig, A. Lopez-Nogueira, C. Cavaleiro, and N. Pedrol. 2018. On the bioherbicide potential of Ulex europaeus and Cytisus scoparius: Profiles of volatile organic compounds and their phytotoxic effects. PLoS One 13(10):e0205997. Pardo-Muras, M., C. G. Puig, P. Souza-Alonso, and N. Pedrol. 2020. The Phytotoxic Potential of the Flowering Foliage of Gorse (Ulex europaeus) and Scotch Broom (Cytisus scoparius), as Pre-Emergent Weed Control in Maize in a Glasshouse Pot Experiment. Plants 9(2):203. Parsons, W., and E. Cuthbertson. 1992. Noxious Weeds of Australia. Inkata Press. Melbourne/Sydney. Paungfoo-Lonhienne, C., Y. K. Yeoh, N. R. P. Kasinadhuni, T. G. Lonhienne, N. Robinson, P. Hugenholtz, M. A. Ragan, and S. Schmidt. 2015. Nitrogen fertilizer dose alters fungal communities in sugarcane soil and rhizosphere. Scientific Reports 5:8678.

125

Paynter, Q., Y. M. Buckley, P. Peterson, A. Hugh Gourlay, and S. V. Fowler. 2016. Breaking and remaking a seed and seed predator interaction in the introduced range of Scotch Broom (Cytisus scoparius) in New Zealand. Journal of Ecology 104:182-192. Paynter, Q., A. Gourlay, C. Rolando, and M. Watt. 2012. Dispersal of the Scotch broom gall mite Aceria genistae implications for biocontrol. New Zealand Plant Protection 65:81-84. Peltzer, D. A., P. J. Bellingham, H. Kurokawa, L. R. Walker, D. A. Wardle, and G. W. Yeates. 2009. Punching above their weight: low‐biomass non‐native plant species alter soil properties during primary succession. Oikos 118:1001-1014. Peng, W., Y. Zhu, M. Song, H. Du, T. Song, F. Zeng, F. Zhang, K. Wang, Y. Luo, and X. Lan. 2019. The spatial distribution and drivers of soil microbial richness and diversity in a karst broadleaf forest. Forest Ecology and Management 449:117241. Pérez‐Fernández, M., E. Calvo‐Magro, J. Rodríguez‐Sánchez, and A. Valentine. 2017. Differential growth costs and nitrogen fixation in Cytisus multiflorus (L′ Hér.) Sweet and Cytisus scoparius (L.) Link are mediated by sources of inorganic N. Plant Biology 19:742- 748. Pernilla Brinkman, E., W. H. Van der Putten, E. J. Bakker, and K. J. Verhoeven. 2010. Plant–soil feedback: experimental approaches, statistical analyses and ecological interpretations. Journal of Ecology 98:1063-1073. Pesaro, M., F. Widmer, G. Nicollier, and J. Zeyer. 2003. Effects of freeze–thaw stress during soil storage on microbial communities and methidathion degradation. Soil Biology and Biochemistry 35:1049-1061. Peterson, D. J., and R. Prasad. 1998. The biology of Canadian weeds. 109. Cytisus scoparius (L.) Link. Canadian Journal of Plant Science 78:497-504. Pickett, B., I. C. Irvine, E. Bullock, K. Arogyaswamy, and E. Aronson. 2019. Legacy effects of invasive grass impact soil microbes and native shrub growth. Invasive Plant Science and Management 12:22-35. Polz, M. F., and C. M. Cavanaugh. 1998. Bias in template-to-product ratios in multitemplate PCR. Applied and Environmental Microbiology 64:3724-3730. Potter, K., D. J. Kriticos, M. Watt, and A. Leriche. 2009. The current and future potential distribution of Cytisus scoparius: a weed of pastoral systems, natural ecosystems and plantation forestry. Weed Research 49:271-282. Powell, K. I., J. M. Chase, and T. M. Knight. 2011. A synthesis of plant invasion effects on biodiversity across spatial scales. American Journal of Botany 98:539-548. Powers, R. F. 1990. Nitrogen mineralization along an altitudinal gradient: interactions of soil temperature, moisture, and substrate quality. Forest Ecology and Management 30:19-29. Prévosto, B., A. Robert, and P. Coquillard. 2004. Development of Cytisus scoparius L. at stand and individual level in a mid-elevation mountain of the French Massif Central. Acta Oecologica 25:73-81. Prince, A. M., and L. Andrus. 1992. PCR: how to kill unwanted DNA. Biotechniques 12:358-360. Pringle, A., J. D. Bever, M. Gardes, J. L. Parrent, M. C. Rillig, and J. N. Klironomos. 2009. Mycorrhizal symbioses and plant invasions. Annual Review of Ecology, Evolution, and Systematics 40:699-715. Purahong, W., T. Wubet, D. Krüger, and F. Buscot. 2018. Molecular evidence strongly supports deadwood-inhabiting fungi exhibiting unexpected tree species preferences in temperate forests. The ISME Journal 12:289-295. Raaijmakers, J. M., and M. Mazzola. 2016. Soil immune responses. Science 352:1392-1393. Radley, L. 1989. Prehistoric New Zealand. Journal of the Royal Society of New Zealand 19:346- 347. Raghothama, K. 1999. Phosphate acquisition. Annual review of plant biology 50:665-693. Ramirez, K. S., L. B. Snoek, K. Koorem, S. Geisen, L. J. Bloem, F. Ten Hooven, O. Kostenko, N. Krigas, M. Manrubia, and D. Caković. 2019. Range-expansion effects on the belowground plant microbiome. Nature Ecology & Evolution 3:604-611. Ranjard, L., D. P. Lejon, C. Mougel, L. Schehrer, D. Merdinoglu, and R. Chaussod. 2003. Sampling strategy in molecular microbial ecology: influence of soil sample size on DNA

126

fingerprinting analysis of fungal and bacterial communities. Environmental microbiology 5:1111-1120. Rascher, K. G., A. Große-Stoltenberg, C. Máguas, J. A. A. Meira-Neto, and C. Werner. 2011. Acacia longifolia invasion impacts vegetation structure and regeneration dynamics in open dunes and pine forests. Biological Invasions 13:1099-1113. Rawls, W., Y. A. Pachepsky, J. Ritchie, T. Sobecki, and H. Bloodworth. 2003. Effect of soil organic carbon on soil water retention. Geoderma 116:61-76. Rees, M., and Q. Paynter. 1997. Biological control of Scotch broom: modelling the determinants of abundance and the potential impact of introduced insect herbivores. Journal of Applied Ecology 34:1203-1221. Reid, T. 1973. Nitrogen fixation by Ulex europaeus (gorse) and Cytisus scoparius (broom). Lincoln College, University of Canterbury. Reinhart, K. O., and R. M. Callaway. 2006. Soil biota and invasive plants. New Phytologist 170:445-457. Reinhart, K. O., Y. Lekberg, J. Klironomos, and H. Maherali. 2017. Does responsiveness to arbuscular mycorrhizal fungi depend on plant invasive status? Ecology and Evolution 7:6482-6492. Reinhart, K. O., A. Packer, W. H. Van der Putten, and K. Clay. 2003. Plant–soil biota interactions and spatial distribution of black cherry in its native and invasive ranges. Ecology Letters 6:1046-1050. Rejmánek, M., and D. M. Richardson. 2013. Trees and shrubs as invasive alien species–2013 update of the global database. Diversity and Distributions 19:1093-1094. Richardson, A. E., and R. J. Simpson. 2011. Soil microorganisms mediating phosphorus availability update on microbial phosphorus. Plant Physiology 156:989-996. Richardson, B., A. Vanner, J. Ray, and J. Balneaves. 1997. Effect of some common weed species on Pinus radiata growth at a dry South Island site. Pp. 373-376 In Proceedings of the 50th New Zealand Plant Protection Conference. New Zealand Planbt Protectoin Society Richardson, D. M., N. Allsopp, C. M. D'Antonio, S. J. Milton, and M. Rejmánek. 2000. Plant invasions–the role of mutualisms. Biological Reviews 75:65-93. Rillig, M. C., J. Antonovics, T. Caruso, A. Lehmann, J. R. Powell, S. D. Veresoglou, and E. Verbruggen. 2015. Interchange of entire communities: microbial community coalescence. Trends in Ecology & Evolution 30:470-476. Rivero, J., D. Álvarez, V. Flors, C. Azcón‐Aguilar, and M. J. Pozo. 2018. Root metabolic plasticity underlies functional diversity in mycorrhiza‐enhanced stress tolerance in tomato. New Phytologist 220:1322-1336. Rodríguez-Echeverría, S., J. A. Crisóstomo, C. Nabais, and H. Freitas. 2009. Belowground mutualists and the invasive ability of Acacia longifolia in coastal dunes of Portugal. Biological Invasions 11:651-661. Rodríguez-Echeverría, S., and A. Traveset. 2015. Putative linkages between below-and aboveground mutualisms during alien plant invasions. AoB Plants 7:plv062. Rognes, T., T. Flouri, B. Nichols, C. Quince, and F. Mahé. 2016. VSEARCH: a versatile open source tool for metagenomics. PeerJ 4:e2584. Saia, S., E. Aissa, F. Luziatelli, M. Ruzzi, G. Colla, A. G. Ficca, M. Cardarelli, and Y. Rouphael. 2020a. Growth-promoting bacteria and arbuscular mycorrhizal fungi differentially benefit tomato and corn depending upon the supplied form of phosphorus. Mycorrhiza 30:133- 147. Saia, S., E. Tamayo, C. Schillaci, and P. De Vita. 2020b. Arbuscular mycorrhizal fungi and nutrient cycling in cropping systems. Pages 87-115 Carbon and Nitrogen Cycling in Soil. Springer. Sato, H., Y. Sogo, H. Doi, and H. Yamanaka. 2017. Usefulness and limitations of sample pooling for environmental DNA metabarcoding of freshwater fish communities. Scientific Reports 7:14860.

127

Saunders, J., G. Greer, G. Bourdôt, C. Saunders, T. James, C. Rolando, J. Monge, and M. Watt. 2017. The economic costs of weeds on productive land in New Zealand. International Journal of Agricultural Sustainability 15:380-392. Scherber, C., N. Eisenhauer, W. W. Weisser, B. Schmid, W. Voigt, M. Fischer, E.-D. Schulze, C. Roscher, A. Weigelt, and E. Allan. 2010. Bottom-up effects of plant diversity on multitrophic interactions in a biodiversity experiment. Nature 468:553-556. Schlatter, D. C., K. Kahl, B. Carlson, D. R. Huggins, and T. Paulitz. 2018. Fungal community composition and diversity vary with soil depth and landscape position in a no-till wheat- based cropping system. FEMS Microbiology Ecology 94:fiy098. Schoch, C. L., K. A. Seifert, S. Huhndorf, V. Robert, J. L. Spouge, C. A. Levesque, W. Chen, and F. B. Consortium. 2012. Nuclear ribosomal internal transcribed spacer (ITS) region as a universal DNA barcode marker for Fungi. Proceedings of the National Academy of Sciences 109:6241-6246. Schroeder, J. W., A. Dobson, S. A. Mangan, D. F. Petticord, and E. A. Herre. 2020. Mutualist and pathogen traits interact to affect plant community structure in a spatially explicit model. Nature Communications 11:1-10. Schultheis, E. H., and D. J. MacGuigan. 2018. Competitive ability, not tolerance, may explain success of invasive plants over natives. Biological Invasions 20:2793-2806. Semchenko, M., J. W. Leff, Y. M. Lozano, S. Saar, J. Davison, A. Wilkinson, B. G. Jackson, W. J. Pritchard, R. Jonathan, and S. Oakley. 2018. Fungal diversity regulates plant-soil feedbacks in temperate grassland. Science Advances 4:eaau4578. Shaben, J., and J. H. Myers. 2010. Relationships between Scotch broom (Cytisus scoparius), soil nutrients, and plant diversity in the Garry oak savannah ecosystem. Plant Ecology 207:81- 91. Sharp, C. E., A. L. Brady, G. H. Sharp, S. E. Grasby, M. B. Stott, and P. F. Dunfield. 2014. Humboldt’s spa: microbial diversity is controlled by temperature in geothermal environments. The ISME Journal 8:1166-1174. Sharpley, A., and J. Syers. 1979. Effect of aerial topdressing with superphosphate on the loss of phosphate from a pasture catchment. New Zealand Journal of Agricultural Research 22:273-277. Sheppard, A., and J. Hosking. 2000. Broom management. Workshop held at Ellerton and Moonan, Australia, 16-17 November 1998. Plant Protection Quarterly 15:133-186. Sheppard, A. W., P. Hodge, Q. Paynter, and M. Rees. 2002. Factors affecting invasion and persistence of broom Cytisus scoparius in Australia. Journal of Applied Ecology 39:721-734. Siebert, P. D., and J. W. Larrick. 1992. Competitive PCR. Nature 359:557-558. Sielaff, A. C., R. N. Upton, K. S. Hofmockel, X. Xu, H. W. Polley, and B. J. Wilsey. 2018. Microbial community structure and functions differ between native and novel (exotic- dominated) grassland ecosystems in an 8-year experiment. Plant and Soil 432:359-372. Simberloff, D., and B. Von Holle. 1999. Positive interactions of nonindigenous species: invasional meltdown? Biological Invasions 1:21-32. Simonsen, A., Dinnage, R., Barrett, L., Prober, S., Thrall, P. 2017. Symbiosis limits establishment of legumes outside their native range at a global scale. Nature Communications 8: 14790. Singh, M., and W. M. Meyer. 2020. Plant-Soil Feedback Effects on Germination and Growth of Native and Non-Native Species Common across Southern California. Diversity 12(6):217. Slesak, R. A., T. B. Harrington, and A. W. D’Amato. 2016. Invasive Scotch broom alters soil chemical properties in Douglas-fir forests of the Pacific Northwest, USA. Plant and Soil 398:281-289. Smith, S. E., and D. J. Read. 2010. Mycorrhizal symbiosis. Academic press. Smith, V. H. 1992. Effects of nitrogen: phosphorus supply ratios on nitrogen fixation in agricultural and pastoral ecosystems. Biogeochemistry 18:19-35. Sokol, N. W., S. E. Kuebbing, and M. A. Bradford. 2017. Impacts of an invasive plant are fundamentally altered by a co‐occurring forest disturbance. Ecology 98:2133-2144. Song, S., C. Zhang, Y. Gao, X. Zhu, R. Wang, M. Wang, Y. Zheng, L. Hou, M. Liu, and D. Wu. 2020. Responses of wetland soil bacterial community and edaphic factors to two-year

128

experimental warming and Spartina alterniflora invasion in Chongming Island. Journal of Cleaner Production 250:119502. Song, Z., D. Schlatter, P. Kennedy, L. L. Kinkel, H. C. Kistler, N. Nguyen, and S. T. Bates. 2015. Effort versus reward: preparing samples for fungal community characterization in high- throughput sequencing surveys of soils. PLoS One 10(5):e0127234. Sosa-Hernández, M. A., J. Roy, S. Hempel, T. Kautz, U. Köpke, M. Uksa, M. Schloter, T. Caruso, and M. C. Rillig. 2018. Subsoil arbuscular mycorrhizal fungal communities in arable soil differ from those in topsoil. Soil Biology and Biochemistry 117:83-86. Spagnoletti, F., M. Carmona, N. E. T. Gómez, V. Chiocchio, and R. S. Lavado. 2017. Arbuscular mycorrhiza reduces the negative effects of M. phaseolina on soybean plants in arsenic- contaminated soils. Applied Soil Ecology 121:41-47. Spatz, D. R., K. M. Zilliacus, N. D. Holmes, S. H. Butchart, P. Genovesi, G. Ceballos, B. R. Tershy, and D. A. Croll. 2017. Globally threatened vertebrates on islands with invasive species. Science Advances 3:e1603080. St-Arnaud, M., and V. Vujanovic. 2007. Mycorrhizae in crop production. The Haworth Press, Inc, NY. Steens, M. I., D. J. Winter, R. Morris, J. McCartney, and P. Greenslade. 2007. New Zealand's giant Collembola: new information on distribution and morphology for Holacanthella Börner, 1906 (Neanuridae: Uchidanurinae). New Zealand Journal of Zoology 34:63-78. Stefanowicz, A. M., M. Stanek, M. L. Majewska, M. Nobis, and S. Zubek. 2019. Invasive plant species identity affects soil microbial communities in a mesocosm experiment. Applied Soil Ecology 136:168-177. Stouffer, D. B., A. R. Cirtwill, and J. Bascompte. 2014. How exotic plants integrate into pollination networks. Journal of Ecology 102:1442-1450. Suding, K. N., W. Stanley Harpole, T. Fukami, A. Kulmatiski, A. S. MacDougall, C. Stein, and W. H. van der Putten. 2013. Consequences of plant–soil feedbacks in invasion. Journal of Ecology 101:298-308. Sutela, S., A. Poimala, and E. J. Vainio. 2019. Viruses of fungi and oomycetes in the soil environment. FEMS Microbiology Ecology 95:fiz119. Syrett, P. 2000. Status of broom in New Zealand. Plant Protection Quarterly 15(4):148. Syrett, P., S. Fowler, E. Coombs, J. Hosking, G. Markin, Q. Paynter, and A. Sheppard. 1999. The potential for biological control of Scotch broom (Cytisus scoparius) (Fabaceae) and related weedy species. Biocontrol News and Information 20:17N-34N. Syrett, P., S. V. Fowler, H. M. Harman, L. M. Hayes, J. Memmott, and J. J. Sheat. 2007. Establishment of Arytainilla spartiophila Förster (Hemiptera: Psyllidae), a new biological control agent for broom, Cytisus scoparius, in New Zealand. New Zealand Entomologist 30:53-62. Taberlet, P., A. Bonin, E. Coissac, and L. Zinger. 2018. Environmental DNA: For biodiversity research and monitoring. Oxford University Press. Taberlet, P., E. Coissac, F. Pompanon, C. Brochmann, and E. Willerslev. 2012. Towards next‐ generation biodiversity assessment using DNA metabarcoding. Molecular Ecology 21:2045-2050. Taylor, D. L., T. N. Hollingsworth, J. W. McFarland, N. J. Lennon, C. Nusbaum, and R. W. Ruess. 2014. A first comprehensive census of fungi in soil reveals both hyperdiversity and fine‐scale niche partitioning. Ecological Monographs 84:3-20. Team, R. C. 2013. R: A language and environment for statistical computing. Tedersoo, L., and M. Bahram. 2019. Mycorrhizal types differ in ecophysiology and alter plant nutrition and soil processes. Biological Reviews 94:1857-1880. Tedersoo, L., M. Bahram, T. Cajthaml, S. Põlme, I. Hiiesalu, S. Anslan, H. Harend, F. Buegger, K. Pritsch, and J. Koricheva. 2016. Tree diversity and species identity effects on soil fungi, protists and animals are context dependent. The ISME Journal 10:346. Tedersoo, L., M. Bahram, S. Põlme, U. Kõljalg, N. S. Yorou, R. Wijesundera, L. V. Ruiz, A. M. Vasco-Palacios, P. Q. Thu, and A. Suija. 2014. Global diversity and geography of soil fungi. Science 346:1256688.

129

Tedersoo, L., M. Bahram, and M. Zobel. 2020. How mycorrhizal associations drive plant population and community biology. Science 367:eaba1223. Tedersoo, L., U. Kõljalg, N. Hallenberg, and K. H. Larsson. 2003. Fine scale distribution of ectomycorrhizal fungi and roots across substrate layers including coarse woody debris in a mixed forest. New Phytologist 159:153-165. Teste, F. P., M. D. Jones, and I. A. Dickie. 2020. Dual‐mycorrhizal plants: their ecology and relevance. New Phytologist 225:1835-1851. Teste, F. P., P. Kardol, B. L. Turner, D. A. Wardle, G. Zemunik, M. Renton, and E. Laliberté. 2017. Plant-soil feedback and the maintenance of diversity in Mediterranean-climate shrublands. Science 355:173-176. Toole, K., P. Roffey, E. Young, K. Cho, T. Shaw, M. Smith, and N. Blagojevic. 2019. Evaluation of commercial forensic DNA extraction kits for decontamination and extraction of DNA from biological samples contaminated with radionuclides. Forensic Science International 302:109867. Torchin, M. E., and C. E. Mitchell. 2004. Parasites, pathogens, and invasions by plants and animals. Frontiers in Ecology and the Environment 2:183-190. Tran, H., K. C. Harrington, A. W. Robertson, and M. S. Watt. 2016. Assessment of herbicides for selectively controlling broom (Cytisus scoparius) growing with radiata pine (Pinus radiata) in New Zealand. New Zealand Journal of Forestry Science 46:13. Traveset, A., R. Heleno, S. Chamorro, P. Vargas, C. K. McMullen, R. Castro-Urgal, M. Nogales, H. W. Herrera, and J. M. Olesen. 2013. Invaders of pollination networks in the Galápagos Islands: emergence of novel communities. Proceedings of the Royal Society B: Biological Sciences 280:20123040. Tveit, A., R. Schwacke, M. M. Svenning, and T. Urich. 2013. Organic carbon transformations in high-Arctic peat soils: key functions and microorganisms. The ISME Journal 7:299-311. Upton, R. N., A. Checinska Sielaff, K. S. Hofmockel, X. Xu, H. W. Polley, and B. J. Wilsey. 2020. Soil depth and grassland origin cooperatively shape microbial community co‐ occurrence and function. Ecosphere 11:e02973. Vadakattu, G., and J. Paterson. 2006. Free-living bacteria lift soil nitrogen supply. Farming Ahead 169:40. van den Koornhuyse, P., K. Ridgway, I. Watson, A. Fitter, and J. Young. 2003. Co‐existing grass species have distinctive arbuscular mycorrhizal communities. Molecular Ecology 12:3085- 3095. van der Heijden, M. G., and T. R. Horton. 2009. Socialism in soil? The importance of mycorrhizal fungal networks for facilitation in natural ecosystems. Journal of Ecology 97:1139-1150. van der Heijden, M. G., J. N. Klironomos, M. Ursic, P. Moutoglis, R. Streitwolf-Engel, T. Boller, A. Wiemken, and I. R. Sanders. 1998. Mycorrhizal fungal diversity determines plant biodiversity, ecosystem variability and productivity. Nature 396:69-72. van der Putten, W., G. Kowalchuk, E. Brinkman, G. Doodeman, R. Van der Kaaij, A. Kamp, F. Menting, and E. Veenendaal. 2007a. Soil feedback of exotic savanna grass relates to pathogen absence and mycorrhizal selectivity. Ecology 88:978-988. van der Putten, W., C. Van Dijk, and B. Peters. 1993. Plant-specific soil-borne diseases contribute to succession in foredune vegetation. Nature 362:53-56. van der Putten, W., G. Yeates, H. Duyts, C. S. Reis, and G. Karssen. 2005. Invasive plants and their escape from root herbivory: a worldwide comparison of the root-feeding nematode communities of the dune grass Ammophila arenaria in natural and introduced ranges. Biological Invasions 7:733-746. van der Putten, W. H., R. D. Bardgett, J. D. Bever, T. M. Bezemer, B. B. Casper, T. Fukami, P. Kardol, J. N. Klironomos, A. Kulmatiski, and J. A. Schweitzer. 2013. Plant–soil feedbacks: the past, the present and future challenges. Journal of Ecology 101:265-276. van der Putten, W. H., M. A. Bradford, E. Pernilla Brinkman, T. F. van de Voorde, and G. Veen. 2016. Where, when and how plant–soil feedback matters in a changing world. Functional Ecology 30:1109-1121.

130 van der Putten, W. H., J. N. Klironomos, and D. A. Wardle. 2007b. Microbial ecology of biological invasions. The ISME Journal 1:28-37. van Diepen, L. T., E. A. Lilleskov, and K. S. Pregitzer. 2011. Simulated nitrogen deposition affects community structure of arbuscular mycorrhizal fungi in northern hardwood forests. Molecular Ecology 20:799-811. van Kleunen, M., O. Bossdorf, and W. Dawson. 2018. The ecology and evolution of alien plants. Annual Review of Ecology, Evolution, and Systematics 49:25-47. Vesty, A., K. Biswas, M. W. Taylor, K. Gear, and R. G. Douglas. 2017. Evaluating the impact of DNA extraction method on the representation of human oral bacterial and fungal communities. PLoS One 12(1):e0169877. Vilà, M., J. L. Espinar, M. Hejda, P. E. Hulme, V. Jarošík, J. L. Maron, J. Pergl, U. Schaffner, Y. Sun, and P. Pyšek. 2011. Ecological impacts of invasive alien plants: a meta‐analysis of their effects on species, communities and ecosystems. Ecology Letters 14:702-708. Vitousek, P. M., S. Porder, B. Z. Houlton, and O. A. Chadwick. 2010. Terrestrial phosphorus limitation: mechanisms, implications, and nitrogen–phosphorus interactions. Ecological Applications 20:5-15. Vogelsang, K. M., H. L. Reynolds, and J. D. Bever. 2006. Mycorrhizal fungal identity and richness determine the diversity and productivity of a tallgrass prairie system. New Phytologist 172:554-562. Walker, C., A. Gollotte, and D. Redecker. 2018. A new genus, Planticonsortium (Mucoromycotina), and new combination (P. tenue), for the fine root endophyte, Glomus tenue (basionym Rhizophagus tenuis). Mycorrhiza 28:213-219. Waller, L. P., W. J. Allen, B. Barratt, L. Condron, F. França, J. Hunt, N. Koele, K. H. Orwin, G. Steel, and J. M. Tylianakis. 2020. Biotic interactions drive ecosystem responses to exotic plant invaders. Science 368:967-972. Waloff, N., and O. Richards. 1977. The effect of insect fauna on growth mortality and natality of broom, Sarothamnus scoparius. Journal of Applied Ecology 14:787-798. Wang, F. Y., J. L. Hu, X. G. Lin, S. W. Qin, and J. H. Wang. 2011. Arbuscular mycorrhizal fungal community structure and diversity in response to long-term fertilization: a field case from China. World Journal of Microbiology and Biotechnology 27:67-74. Wang, X., W. Ding, and H. Lambers. 2019. Nodulation promotes cluster-root formation in Lupinus albus under low phosphorus conditions. Plant and Soil 439:233-242. Wardle, D. A. 2013. Communities and ecosystems: linking the aboveground and belowground components (MPB-34). Princeton University Press. Warrington, S., Ellis, A., Novoa, A., Wandrag, E.M., Hulme, P.E., Duncan, R.P., Valentine, A. and Le Roux, J.J., 2019. Cointroductions of Australian acacias and their rhizobial mutualists in the Southern Hemisphere. Journal of Biogeography, 46(7), pp.1519-1531. Waterhouse, B. 1988. Broom (Cytisus scoparius) at Barrington Tops, New South Wales. Australian Geographical Studies 26:239-248. Watt, M. S., P. W. Clinton, D. Whitehead, B. Richardson, E. G. Mason, and A. C. Leckie. 2003a. Above-ground biomass accumulation and nitrogen fixation of broom (Cytisus scoparius L.) growing with juvenile Pinus radiata on a dryland site. Forest Ecology and Management 184:93-104. Watt, M. S., D. Whitehead, E. G. Mason, B. Richardson, and M. O. Kimberley. 2003b. The influence of weed competition for light and water on growth and dry matter partitioning of young Pinus radiata, at a dryland site. Forest Ecology and Management 183:363-376. Wearne, L. J., and J. W. Morgan. 2004. Community‐level changes in Australian subalpine vegetation following invasion by the non‐native shrub Cytisus scoparius. Journal of Vegetation Science 15:595-604. Webb, C. J., W. R. Sykes, and P. J. Garnock-Jones. 1988. Flora of New Zealand. Department of Scientific and Industrial Research (New Zealand), Botany Division. Webster, J., and R. Weber. 2007. Introduction to fungi. Cambridge University Press. Weigelt, A., R. Bol, and R. D. Bardgett. 2005. Preferential uptake of soil nitrogen forms by grassland plant species. Oecologia 142:627-635.

131

Weir, B. S., S. J. Turner, W. B. Silvester, D.-C. Park, and J. M. Young. 2004. Unexpectedly diverse Mesorhizobium strains and Rhizobium leguminosarum nodulate native legume genera of New Zealand, while introduced legume weeds are nodulated by Bradyrhizobium species. Applied and Environmental Microbiology 70:5980-5987. White, T. J., T. Bruns, S. Lee, and J. Taylor. 1990. Amplification and direct sequencing of fungal ribosomal RNA genes for phylogenetics. Pp. 315-322 In: PCR protocols: a guide to methods and applications. Academic Press, Inc., NY. Whittaker, R. H. 1970. Communities and ecosystems. New York: Macmillan Company. Wickham, H., Francois, R., Henry, L., & Müller, K. (2015). dplyr: A grammar of data manipulation. R package version 0.4, 3. Will, G., G. Will, C. M. Will, G. Will, and V. Juarez. 1985. Nutrient deficiencies and fertiliser use in New Zealand exotic forests. Forest Research Institute, New Zealand Forest Service. Williams, P. 1981. Aspects of the ecology of broom (Cytisus scoparius) in Canterbury, New Zealand. New Zealand Journal of Botany 19:31-43. Williams, P. A., J. M. Kean, and R. P. Buxton. 2010. Multiple factors determine the rate of increase of an invading non-native tree in New Zealand. Biological Invasions 12:1377- 1388. Williamson, G. B. 1990. Allelopathy, Koch’s postulates, and the neck riddle. Perspectives on plant competition. Academic Press, Inc. California, USA. pp. 143-162. Wilschut, R. A., W. H. van der Putten, P. Garbeva, P. Harkes, W. Konings, P. Kulkarni, H. Martens, and S. Geisen. 2019. Root traits and belowground herbivores relate to plant–soil feedback variation among congeners. Nature Communications 10:1-9. Wilson, H., T. McDonald, and D. Lamb. 2017. Forest regeneration on Hinewai Reserve, New Zealand: An interview with Hugh Wilson. Ecological Management & Restoration 18:92- 102. Wilson, I. G. 1997. Inhibition and facilitation of nucleic acid amplification. Applied and Environmental Microbiology 63:3741. Wilson, J. B. 1988. A review of evidence on the control of shoot: root ratio, in relation to models. Annals of Botany 61:433-449. Wood, C. L., J. E. Byers, K. L. Cottingham, I. Altman, M. J. Donahue, and A. M. Blakeslee. 2007. Parasites alter community structure. Proceedings of the National Academy of Sciences 104:9335-9339. Wright, S. H., S. M. Berch, and M. L. Berbee. 2009. The effect of fertilization on the below-ground diversity and community composition of ectomycorrhizal fungi associated with western hemlock (Tsuga heterophylla). Mycorrhiza 19:267-276. Wubs, E. J., W. H. Van der Putten, M. Bosch, and T. M. Bezemer. 2016. Soil inoculation steers restoration of terrestrial ecosystems. Nature Plants 2:1-5. Xiao, H. F., Y. L. Feng, D. A. Schaefer, and X. D. Yang. 2014. Soil fungi rather than bacteria were modified by invasive plants, and that benefited invasive plant growth. Plant and Soil 378:253-264. Xiao, X., Y. Liang, S. Zhou, S. Zhuang, and B. Sun. 2018. Fungal community reveals less dispersal limitation and potentially more connected network than that of bacteria in bamboo forest soils. Molecular Ecology 27:550-563. Yan, D., J. G. Mills, N. J. Gellie, A. Bissett, A. J. Lowe, and M. F. Breed. 2018. High-throughput eDNA monitoring of fungi to track functional recovery in ecological restoration. Biological Conservation 217:113-120. Yang, Y., Y. Dou, Y. Huang, and S. An. 2017. Links between soil fungal diversity and plant and soil properties on the Loess Plateau. Frontiers in Microbiology 8:2198. Yelenik, S. G., and J. M. Levine. 2011. The role of plant–soil feedbacks in driving native‐species recovery. Ecology 92:66-74. Zak, J. C., and M. R. Willig. 2004. Fungal biodiversity patterns. Biodiversity of fungi. Inventory and monitoring methods. Elsevier Academic, USA. pp. 59-75. Zanne, A. E., K. Abarenkov, M. E. Afkhami, C. A. Aguilar‐Trigueros, S. Bates, J. M. Bhatnagar, P. E. Busby, N. Christian, W. K. Cornwell, and T. W. Crowther. 2019. Fungal functional

132

ecology: bringing a trait‐based approach to plant‐associated fungi. Biological Reviews 95: 409-433. Zebarth, B., C. Goyer, S. Neupane, S. Li, A. Mills, S. Whitney, A. Cambouris, and I. Perron. 2018. Soil microbial communities have distinct spatial patterns in agricultural fields. In Proceedings of the 14th International Conference on Precision Agriculture. Montreal. Canada. Zenni, R. D., I. A. Dickie, M. J. Wingfield, H. Hirsch, C. J. Crous, L. A. Meyerson, T. I. Burgess, T. G. Zimmermann, M. M. Klock, and E. Siemann. 2017. Evolutionary dynamics of tree invasions: complementing the unified framework for biological invasions. AoB Plants 9: plw085. Zenni, R. D., and M. A. Nuñez. 2013. The elephant in the room: the role of failed invasions in understanding invasion biology. Oikos 122:801-815. Zhang, F.-J., Q. Li, F.-X. Chen, H.-Y. Xu, and F.-H. Wan. 2017. Arbuscular mycorrhizal fungi facilitate growth and competitive ability of an exotic species Flaveria bidentis. Soil Biology and Biochemistry 115:275-284. Zhang, F., Q. Li, E. H. Yerger, X. Chen, Q. Shi, and F. Wan. 2018. AM fungi facilitate the competitive growth of two invasive plant species, Ambrosia artemisiifolia and Bidens pilosa. Mycorrhiza 28:703-715. Zhou, J., X. Jiang, B. Zhou, B. Zhao, M. Ma, D. Guan, J. Li, S. Chen, F. Cao, and D. Shen. 2016. Thirty four years of nitrogen fertilization decreases fungal diversity and alters fungal community composition in black soil in northeast China. Soil Biology and Biochemistry 95:135-143. Zhu, Q., W. Riley, J. Tang, and C. Koven. 2016. Multiple soil nutrient competition between plants, microbes, and mineral surfaces: model development, parameterization, and example applications in several tropical forests. Biogeosciences 13:341-363. Zinger, L., A. Bonin, I. G. Alsos, M. Bálint, H. Bik, F. Boyer, A. A. Chariton, S. Creer, E. Coissac, and B. E. Deagle. 2019. DNA metabarcoding—Need for robust experimental designs to draw sound ecological conclusions. Molecular Ecology 28:1857-1862.

133

Appendix A (Supplement Chapter 2)

A1. Coordinates of 18 marked permanent sampling plots in Molesworth used for soil collection. All plots are within 2.5 km of each other at an altitude 872-933 m above sea level. The plots were set up by Manaaki Whenua and follow field protocols outlined by Hurst and Allen (1993).

Plot East Longitude North Latitude MW1 24°96'108'' 58°60'748'' MW2 24°96'131'' 58°60'767'' MW3 24°96'171'' 58°60'777'' MW6 24°96'121'' 58°60'592'' MW7 24°96'095'' 58°60'563'' MW12 24°96'348'' 58°60'808'' MW13 24°96'011'' 58°60'642'' MW14 24°96'051'' 58°60'651'' MW15 24°95'971'' 58°60'651'' MW17 24°95'592'' 58°60'764'' MW18 24°95'673'' 58°60'782'' MW19 24°95'651'' 58°60'673'' MW20 24°95'630'' 58°60'779'' MW23 24°95'753'' 58°60'794'' MW24 24°95'769'' 58°60'824'' MW25 24°95'788'' 58°60'780'' MW26 24°95'822'' 58°60'755'' MW29 24°95'855'' 58°60'681''

A2. [Corresponds to Table 3] Linear mixed-effect model t-values for C. scoparius coverage × plant origin × legume status (top table) and C. scoparius coverage × plant origin × ectomycorrhizal (ECM) status (bottom table). Non-significant terms only included if part of significant higher level interaction. “.” indicates term dropped during model simplification.

Broom coverage Broom coverage Fabaceae Broom coverage Broom coverage Fabaceae Native × Fabaceae × Native × Native × Fabaceae × Native

Shoot mass (g) 1.024 1.070 0.228 6.629 2.112 -0.919 -5.014 Root mass (g) 0.267 -0.282 0.474 3.848 2.006 -0.464 -3.158 Whole plant mass (g) 0.308 0.992 0.485 5.862 1.699 -1.039 -3.992 Shoot N (%) 2.461 4.875 0.058 -2.402 -1.709 -1.643 2.464 Shoot P (%) . -1.743 -0.850 . . 2.119 . Shoot N (%) / P (%) 1.986 3.989 0.215 . -3.320 -1.962 . Total N / Total P 0.746 1.749 0.148 7.241 0.924 -1.105 -4.907

Broom coverage Broom coverage ECM Broom coverage Broom coverage ECM Native × ECM × Native × Native × ECM × Native

Shoot mass (g) 6.506 -0.633 -0.901 -4.282 -3.376 0.918 3.275 Root mass (g) 4.700 0.345 -0.655 -2.438 . 1.989 . Whole plant mass (g) 6.717 -0.233 -0.470 -4.433 -3.538 0.966 2.281 Shoot N (%) 0.425 -2.756 . 2.680 . . . Shoot P (%) ...... Shoot N (%) / P (%) 0.740 -1.635 -1.407 3.544 -2.769 . . Total N / Total P 7.057 -0.862 -1.033 -4.031 -4.331 0.664 2.848

134

A3. [Corresponds to Table 3] Linear mixed-effect model estimates for log-transformed distance to closest mature C. scoparius × plant origin × legume status. t-values are outside of parentheses, P- values within parentheses. Non-significant terms only included if part of significant higher level interaction. “.” indicates term dropped during model simplification. Underlined values show which estimates differ in significance between the estimates in Table 3.

Distance to broom Distance to broom Fabaceae Distance to broom Distance to broom Fabaceae Native × Fabaceae × Native × Native × Fabaceae × Native

Shoot mass (g) -0.888 (< 0.0001) 7.441 (0.0339) 2.113 (0.1774) -6.622 (0.0003) -1.884 (0.0201) -5.797 (0.0068) 5.127 (< 0.0001) Root mass (g) -0.329 (< 0.0001) 4.367 (0.9352) 2.516 (0.9739) -4.122 (0.0525) -2.155 (0.3774) -4.230 (0.0549) 3.795 (0.0012) Whole plant mass (g) -0.281 (< 0.0001) 7.094 (0.0349) 2.131 (0.2761) -6.275 (< 0.0001) -1.818 (0.0190) -5.378 (0.0084) 4.711 (< 0.0001) Shoot N (%) -2.143 (0.1594) 0.222 (< 0.0001) -1.591 (0.1179) 2.027 (0.5506) 1.547 (0.9824) 1.640 (0.3720) -2.305 (0.0211) Shoot P (%) . -1.743 (0.7259) -0.850 (0.3789) . . 2.119 (0.0341) . Shoot N (%) / P (%) -2.218 (0.4828) 3.310 (0.0009) -4.046 (0.0155) . 3.258 (0.0011) . . Total N / Total P -0.647 (< 0.0001) 8.479 (0.0005) 0.926 (0.0156) -7.187 (< 0.0001) -0.790 (0.0007) -5.873 (0.0017) 5.025 (< 0.0001)

135

A4. [Corresponds to Figure 4] Standardised and centered root biomass, whole plant biomass and height at harvest for each plant over C. scoparius % coverage. Regression lines are shown when P < 0.05.

136

A4. [Continued]

A5. [Corresponds to Figure 4] Standardised and centered aboveground biomass for each plant over distance to the closest mature C. scoparius. Regression lines are shown when P < 0.05.

137

A6. Soil nutrients over C. scoparius % coverage. P values and adjusted R2 are given in the plots.

138

Appendix B (Supplement Chapter 3)

B1. Pie chart of 5263 fungal OTUs according to fungal taxa. The most abundant member of Ascomycota, Basidiomycota, Mortierellomycotina, Glomeromycotina, Mucoromycotina and Chytridiomycotina were a species of Lecanicillium, Serendipita, Mortierella, Archaeosporaceae (within the family of Glomales), Umbelopsis and Phlyctochytrium respectively.

139

B2. Gamma (γ) diversity of fungal OTUs according to fungal taxa and plot. Plots are in ascending order of mean C. scoparius coverage and the first three plots (MW13, MW19 and MW26) contain no C. scoparius. Note that gamma diversity is not presented as integers because the mean rarefied gamma diversity was calculated across 250 iterations.

Asco- Basidio- Glomero- Mortierello- Chytridio- Mucoro- Other γ diversity Plot mycota mycota mycotina mycotina mycotina mycotina fungal taxa per plot

MW13 763.1 270.4 10.8 39.1 7.2 14.4 19.2 1124.1 MW19 946.7 403.3 21.1 43.9 7.9 23.9 13.8 1460.7 MW26 648.8 293.2 9.0 58.8 8.7 13.8 13.6 1045.8 MW7 580.7 255.9 4.3 49.6 1.0 12.4 6.3 910.2 MW6 588.2 258.1 4.8 50.8 3.3 20.0 17.4 942.6 MW1 641.3 265.0 8.0 47.2 5.3 17.7 16.1 1000.6 MW23 618.7 243.6 7.6 48.1 5.1 18.0 9.2 950.3 MW18 512.7 171.5 5.2 40.0 3.7 16.7 9.9 759.7 MW15 753.5 279.7 9.3 34.3 13.6 20.6 23.6 1134.6 MW12 636.4 284.7 4.8 51.3 5.2 19.8 11.3 1013.4 MW3 926.4 348.4 16.9 53.9 15.6 20.7 28.9 1410.8 MW14 701.1 271.7 22.5 48.6 9.7 21.5 16.9 1092.1 MW2 605.0 325.1 36.3 42.0 10.6 20.8 15.4 1055.3 MW25 484.4 269.6 26.9 49.4 5.6 24.4 11.0 871.3 MW17 754.8 328.5 43.4 40.5 8.0 34.1 19.8 1229.2 MW20 714.9 305.4 14.7 50.1 5.3 26.4 13.5 1130.4 MW29 800.8 376.5 43.1 51.4 7.9 24.8 13.4 1317.8 MW24 676.0 298.3 26.7 49.7 10.4 29.1 21.3 1111.6

140

B3. [Corresponds to Figure 3] Gamma (γ) diversity of Glomeromycotina (above) and Mucoromycotina (below) over distance from the extracted soil core to the closest mature C. scoparius (mm) (note log-transformed axis). P and R2 values are given in the plots.

141

B4. [Corresponds to Table 1] Linear mixed-effect model t value estimates for alpha diversity (at the level of soil cores) and different measurements of C. scoparius density, i.e., C. scoparius % coverage and log-transformed distance to closest C. scoparius (mm). t value estimates are likewise given for proportional abundance and different measurements of C. scoparius density, except for all fungi (indicated by “.”).

Alpha Diversity Proportional Abundance Broom % log(Distance Broom % log(Distance Coverage to Broom) Coverage to Broom) All fungi 2.289 -3.025 . . Ascomycota 0.587 -2.335 -2.871 0.714 Basidiomycota 3.917 -3.211 1.544 -1.210 Glomeromycotina 1.495 0.065 -1.349 0.554 Mortierellomycotina 3.432 0.661 2.638 0.631 Chytridiomycotina 3.880 -2.919 1.505 -2.814 Mucoromycotina 0.690 -0.589 1.260 -0.326 Antagonists 3.043 -1.901 0.936 0.788 Symbiotrophs 0.103 -1.859 -2.157 0.551 Saprotrophs 1.992 -2.867 0.962 -0.465 Plant pathogens 3.014 -1.711 2.767 -3.094 Pathogens of fungi -0.899 -0.682 1.691 -0.303 ECM (FUNGuild) -0.353 -1.492 -0.795 0.040

142

B5. [Corresponds to Table 1] Linear mixed-effect model estimates for alpha diversity (at the level of soil cores) and proportional abundance and different measurements of C. scoparius density, i.e., C. scoparius % coverage and log-transformed distance to closest C. scoparius (mm). P values are given within parenthesis, t value outside of parentheses.

Alpha Diversity Proportional Abundance Broom % log(Distance Broom % log(Distance Coverage to Broom) Coverage to Broom)

NOT Ascomycota 4.662 (< 0.0001) -2.825 (0.0047) 2.871 (0.0040) -0.714 (0.4754) NOT Basidiomycota 1.431 (0.1523) -2.547 (0.0108) -1.544 (0.1227) 1.210 (0.2262) NOT Glomeromycotina 2.161 (0.0306) -2.938 (0.0033) 1.349 (0.1773) -0.554 (0.5797) NOT Mortierellomycotina 1.810 (0.0702) -2.952 (0.0031) -2.638 (0.0083) -0.631 (0.5281) NOT Chytridiomycotina 2.233 (0.0255) -2.990 (0.0027) -1.505 (0.1324) 2.814 (0.0048) NOT Mucoromycotina 2.233 (0.0255) -3.004 (0.0026) -1.505 (0.1324) 2.814 (0.0048) Antagonists (Loose) 2.505 (0.0122) -3.455 (0.0005) 2.893 (0.0038) -1.269 (0.2046) Symbiotrophs (Loose) 1.963 (0.0496) -2.284 (0.0223) -1.371 (0.1704) 0.444 (0.6569) Saprotrophs (Loose) 2.391 (0.0168) -2.902 (0.0037) 0.413 (0.6799) -0.153 (0.8786) NOT Antagonists (Strict) 2.125 (0.0336) -3.006 (0.0026) -0.936 (0.3491) -0.788 (0.4306) NOT Symbiotrophs (Strict) 2.623 (0.0087) -3.102 (0.0019) 2.157 (0.0309) -0.551 (0.5817) NOT Saprotrophs (Strict) 2.224 (0.0261) -2.831 (0.0046) -0.962 (0.3360) 0.465 (0.6416) NOT Antagonists (Loose) 1.955 (0.0506) -2.362 (0.0182) -2.893 (0.0038) 1.269 (0.2046) NOT Symbiotrophs (Loose) 2.418 (0.0156) -3.133 (0.0017) 1.371 (0.1704) -0.444 (0.6569) NOT Saprotrophs (Loose) 1.557 (0.1194) -2.518 (0.0118) -0.413 (0.6799) 0.153 (0.8786) ECM (FUNGuild) (Loose) -1.089 (0.2763) 0.271 (0.7867) -2.326 (0.0200) 1.163 (0.2446)

143

B6. [Corresponds to Figure 4] Mean alpha diversity per plot of all fungal OTUs over log-transformed distance from the extracted soil core to the stem of the closest C. scoparius (whether mature or immature). P and R2 values are given in the plots.

B7. [Next page; corresponds to Figure 5] Average alpha diversity (at the level of plots) and proportional abundance of fungal OTUs according to fungal taxa and functional traits over distance from the extracted soil core to the base of the closest mature C. scoparius (mm) (note log-transformed axis). Regression lines are shown when P < 0.05. OTUs were ‘strictly’ classified into functional guilds (results with ‘loose’ classifications, i.e., with overlap between functional guilds, are presented in Appendix C8). P and R2 values are given in the plots.

144

145

B8. [Corresponds to Figure 5] Average alpha diversity (at the level of plots) and proportional abundance of fungal OTUs grouped ‘loosely’ (i.e., with overlap) according to functional traits over C. scoparius coverage (below) and over log-transformed distance from the extracted soil core to the base of the closest mature C. scoparius (next page). Regression lines are shown when P < 0.05. OTUs for arbuscular mycorrhizal fungi (AMF) according to FUNGuild (Nguyen et al. 2016), matched exactly with the OTUs for Glomeromycotina according to the UNITE public database (Nilsson et al. 2018). P and R2 values are given in the plots.

146

B8. [Continued]

147

B9. [Corresponds to Figure 6] Beta diversity of Glomeromycotina and Basidiomycota (below) and Mucoromycotina and plant pathogens (next page) over log-transformed distance from extracted soil core to closest mature C. scoparius (mm). P and R2 values are given in the plots.

148

B9. [Continued]

149

B10. [Corresponds to Figure 7] Number of unique Glomeromycotina and Mucoromycotina OTUs occuring in less than half of all plots over C. scoparius % coverage per plot. P and R2 values are given in the plots.

150

B11. OTU occupancy when present (coloured according to fungal phyla) over number of plots in which the OTU has been detected. Note log-transformed x and y axes. The number of OTUs present accoruding to the number of plots is given above each column. There were 41.3% more OTUs (n = 167) found across all three plots with highest C. scoparius coverage compared to across all three plots with lowest C. scoparius coverage. The three plots with highest C. scoparius coverage (below) were MW20, MW24 & MW29. The three plots without C. scoparius coverage (next page) were MW13, MW19 & MW26.

3 plots without C. scoparius coverage

151

B11. [Continued]

3 plots with highest C. scoparius coverage

152

B12. OTU occupancy when present for all fungal taxa and for individual fungal taxa over number of plots in which the OTU has been detected. Note log-transformed x and y axes.

153

B12. [Continued]

154

B12. [Continued]

155

B12. [Continued]

156

Appendix C (Supplement Chapter 4)

C1. All fungal OTUs for both stand-alone and pooled samples split according to taxonomic phyla. Current taxonomy places Glomeromycotina into Mucoromycota, yet due to database limitations, Glomeromycotina follows its older taxonomy of Glomeromycota.

C2. Mean number of fungal OTUs (not rarefied) for each fungal phyla according to the number of samples in a pool. eDNA samples which had less OTUs than stand-alone samples are highlighted in blue and eDNA samples with more OTUs than standaone samples are in green.

Both pooled eDNA samples pooled by and stand-alone Stand-alone Pooled eDNA samples eDNA samples eDNA samples 3 6 12 24 Number of samples 233 143(*) 90 48 24 12 6 All fungal phyla 208.219 206.615 210.767 186.250 220.792 257.750 272.833 Ascomycota 134.622 140.406 125.433 110.750 133.750 151.500 157.500 Basidiomycota 46.279 40.517 55.433 47.125 56.708 72.917 81.833 Glomeromycota 1.927 2.147 1.578 1.542 1.417 1.917 1.833 Mortierellomycotina 17.717 15.573 21.122 20.083 21.417 23.500 23.500

fungal OTUs fungal Chytridiomycotina 0.760 0.797 0.700 0.750 0.625 0.583 0.833 Mean number of number Mean Mucoromycotina 5.236 5.287 5.156 4.958 5.417 5.250 5.500

(*) 144 stand-alone eDNA samples underwent sequencing, yet one was ommitted due to having < 1000 reads

157

C3. [Corresponds to Figures 3 & 4] Linear mixed-effect model estimates for rarefied richness by fungal phylum over log-transformed number of samples per pool (t-values outside of parentheses, p-values within parentheses).

Pooled × Pooled log(samples in pool) log(samples in pool) Rarefied richness of all fungal phyla -3.16 (< 0.0001) 28.541 (< 0.0001) -4.568 (< 0.0001) Rarefied richness of Ascomycota -4.291 (< 0.0001) 25.862 (< 0.0001) -5.368 (< 0.0001) Rarefied richness of Basidiomycota -2.929 (< 0.0001) 40.855 (< 0.0001) -9.345 (< 0.0001) Rarefied richness of Mortierellomycotina 1.94 (< 0.0001) 30.007 (< 0.0001) -9.531 (< 0.0001)

C4. [Corresponds to Figure 5] Linear mixed-effect model estimates for proportional abundance by fungal phylum over log-transformed number of samples per pool (t-values outside of parentheses, p-values within parentheses).

Pooled × Pooled log(samples in pool) log(samples in pool) Proportional abundance of Ascomycota -10.605 (< 0.0001) -0.017 (0.7399) 0.444 (0.6569) Proportional abundance of Basidiomycota 4.646 (< 0.0001) 0.011 (0.8601) -0.24 (0.8107) Proportional abundance of Mortierellomycotina 9.619 (< 0.0001) 0.009 (0.8629) -0.232 (0.8167)

C5. [Corresponds to Figure 7] Linear mixed-effect model estimates for rarefied richness over mean C. scoparius % coverage for all fungal phyla and for individual fungal phylum (t-values outside of parentheses, p-values within parentheses).

Broom Samples in Broom coverage × Samples in pool × coverage pool Pooled samples in pool pooled Rarefied richness of all fungal phyla 1.821 (0.02558) 12.488 (< 0.0001) 0.469 (< 0.0001) 2.188 (0.0287) -7.564 (< 0.0001)

Rarefied richness of Ascomycota 1.852 (0.064) 17.848 (< 0.0001) - 2.324 (< 0.0001) • -7.781 (< 0.0001) Rarefied richness of Basidiomycota 4.903 (0.018343) 17.129 (< 0.0001) 0.754 (< 0.0001) 2.583 (0.009808) -11.902 (< 0.0001) Rarefied richness of Mortierellomycotina 0.08 (0.44781) 11.939 (< 0.0001) 3.539 (< 0.0001) 2.418 (0.01559) -10.076 (< 0.0001)

C6. [Corresponds to Figure 8] Linear mixed-effect model estimates for rarefied richness of individual fungal phyla in proportion to the rarefied richness of all fungal phyla over mean C. scoparius % (t- values outside of parentheses, p-values within parentheses).

Rarefied richness of Broom Samples Broom coverage × Broom coverage × Samples in pool × Broom coverage × coverage in pool Pooled samples in pool pooled pooled samples in pool × pooled Basidiomycota 1.08 (0.294) 12.262 (< 0.0001) 0.04 (< 0.0001) -2.288 (0.205) 0.662 (0.00382) -6.632 (< 0.0001) 2.083 (0.0372) ~ of all fungal phyla Mortierellomycotina -2.317 (0.04234) 0.899 (0.80651) 4.755 (0.01272) 1.683 (0.09242) . -4.281 (< 0.0001) . ~ all fungal phyla Ascomycota 0.091 (0.9277) 1.619 (0.1055) -17.759 (< 0.0001) . . . . ~ all fungal phyla

158

C7. Flowchart for calculated proportional abundance without rarefaction.

159

C8. Taxonomic composition of OTUs from across all plots (below) and for each individual plot (next page), split by the top 3rd, middle 3rd and bottom 3rd proportional rank abundance percentile for each degree of pooling.

160

C8. [Continued]

161

C8. [Continued]

162

C8. [Continued]

163

C8. [Continued]

164

C8. [Continued]

165

C8. [Continued]

166

C9. Proportional rank sequence abundance across all plots for each degree of pooling (3, 6, 12 & 24). The y-axis of the stand-alone samples (in red) is set to double that of the rest.

167

C10. Proportional rank sequence abundance for each plot for each degree of pooling (3, 6, 12 & 24).

168

C10. [Continued]

169

C10. [Continued]

170

C10. [Continued]

171

C10. [Continued]

172

C10. [Continued]

173

C11. Average % occupancy when present of different fungal species coloured by phylum for 24 cores from each sampling plot.

174

C11. [Continued]

175

C11. [Continued]

176

C11. [Continued]

177

C11. [Continued]

178

C11. [Continued]

179

Appendix D (Supplement Chapter 5)

D1. Standardised and centred aboveground dry biomass (g) over C. scoparius (i.e., broom) coverage (above) and distance to closest mature C. scoparius (below). Size of points is scaled according to the square root of AMF rarefied richness and colour ranges from low AMF proportional abundance (blue) to high AMF proportional abundance (red). Regression lines are shown when P < 0.05.

180

D2. Shoot % N over rarefied richness of saprotrophs (i.e., decomposers). Size of points is scaled according to the square root of C. scoparius coverage.

181

D3. Shoot % P over rarefied richness of saprotrophs (above) and over rarefied richness of plant pathogens (below). Size of points is scaled according to the square root of C. scoparius coverage.

182

Appendix E – Allen et al. (2020)

183

Received: 1 September 2019 | Accepted: 3 May 2020 DOI: 10.1111/1365-2745.13433

RESEARCH ARTICLE

Community-level direct and indirect impacts of an invasive plant favour exotic over native species

1The Bio-Protection Research Centre, Lincoln University, Lincoln, New Zealand Abstract 2The Bio-Protection Research Centre, 1. Indirect interactions mediated by shared enemies or mutualists (i.e. apparent School of Biological Sciences, University of competition) can influence whether invasive plants harm or benefit co-occurring Canterbury, Christchurch, New Zealand 3AgResearch, Invermay, Mosgiel, species. However, studies to date have largely examined single pairwise interac- New Zealand tions, limiting our understanding of the interplay among different types of interac-

Correspondence tions and whether indirect impacts systematically favour native or exotic species. Warwick J. Allen Predicting indirect interaction strength has also proven challenging, and it remains Email: [email protected] unclear whether the strengths of different indirect interactions are correlated. Funding information 2. We conducted a field experiment in a grassland invaded by Scotch broom Cytisus Tertiary Education Commission; Royal Society of New Zealand scoparius to compare the strength of its indirect impacts, via both soil fungi and herbivores, on 21 native and exotic legume species growing in pots buried in the Handling Editor: James Cahill ground. Direct interactions of plants with soil fungi were controlled using nylon mesh pot windows of differing porosity (1 or 38 µm) to prevent or allow soil fungi hyphal growth. Arthropod herbivores were controlled through spraying pyrethrum pesticide. To assess indirect impacts, interactions were compared between plants adjacent to or 50 m away from an extensive Scotch broom invasion. We measured plant performance (survival, height and biomass), arthropod and hare herbivory, and rhizobia nodulation. 3. Despite increasing arthropod herbivory of both native and exotic plant species, Scotch broom had a net positive impact on their survival and growth, through sheltering them from abiotic stress, and indirectly via beneficial soil fungi and release from hare browsing. Soil fungi also increased arthropod herbivory, decreased rhizobia nodulation and disproportionately promoted the growth of exotic plants. Overall, exotic plants experienced stronger interactions, which favoured them with beneficial soil fungi and rhizobia but not hare browsing. Finally, indirect interaction strength was not correlated among indirect interactions mediated by different interaction partners. 4. Synthesis. We demonstrate that invaders affect their competitors through multiple interacting indirect pathways that were stronger than direct ‘nurse plant’ effects, emphasizing the importance of a community-level approach to studying biological invasions. Exotic species experienced stronger positive and negative impacts than

Warwick J. Allen and Ralph Wainer contributed equally to this work.

natives, but were facilitated overall, potentially contributing to exotic dominance in communities.

KEYWORDS apparent competition, Cytisus scoparius, indirect facilitation, invasive species, mutualism, nurse plant, plant–fungi interactions, plant–herbivore interactions

1 | INTRODUCTION Another major limitation of existing studies of indirect impacts is that most have been between invasive and native species (Allen Invasive species are exotic species that spread rapidly within a et al., 2018; Bhattarai et al., 2017; Enge et al., 2013), with lim- new range (Richardson et al., 2000), often with positive and neg- ited consideration of indirect interactions between exotic species ative impacts on various aspects of environmental, economic (including those deemed to be invasive). In one of the few pub- and societal well-being (Pimentel, Zuniga, & Morrison, 2005; lished examples, invasive Pinus contorta has been shown to facil- Vilà et al., 2011). Many of their impacts occur through direct in- itate ectomycorrhization of exotic Pseudotsuga menziesii (Dickie teractions of invasive species with other species in the commu- et al., 2014). Moreover, exotic trees tend to share more ecto- nity (Tylianakis, Didham, Bascompte, & Wardle, 2008). However, mycorrhizal fungi than natives (Dickie, Cooper, Bufford, Hulme, indirect interactions, defined as impacts of one species on the & Bates, 2017), suggesting indirect facilitation may be occurring growth, fitness, or population dynamics of another through among exotic plants. However, the impacts of these indirect in- changes in the population or behaviour of a third species, may be teractions on plant performance in native-exotic systems remain as important as direct interactions in influencing invasion success largely unknown. If indirect facilitation occurs between two ex- and impacts (Bhattarai, Meyerson, & Cronin, 2017). Given their otic species via a shared interaction partner, this may represent disproportionately high biomass, abundance (van Kleunen, Weber, an indirect pathway to ‘invasional meltdown’ (Simberloff & Von & Fischer, 2010) and generalist associations (Bartomeus, Vilà, & Holle, 1999), which has received little attention from research- Santamaría, 2008; Moora et al., 2011), invasive species may be ers relative to direct facilitation (Braga, Gómez-Aparicio, Heger, expected to engage in strong indirect interactions with the sur- Vitule, & Jeschke, 2018). On the other hand, apparent competi- rounding community. Yet, the relative importance of direct and in- tion between two exotic species could represent an indirect form direct interactions in modifying invasive species impacts remains of local biotic resistance (‘invasional interference’; Yang, Ferrari, unresolved. & Shea, 2011). Overall, indirect impacts may be expected to be Indirect interactions may be mediated by mutualists or ene- stronger for exotic than native species, due to their generally mies. For example, apparent competition describes negative inter- higher density and biomass (van Kleunen et al., 2010) and ten- actions between two or more species mediated by changes in the dency to be generalist in their species interactions (Bartomeus population or behaviour of shared enemies (Holt, 1977; Holt & et al., 2008; Moora et al., 2011). Bonsall, 2017). Invasive plants can act as a reservoir for harmful her- Indirect interactions among species are highly context depen- bivores (Bhattarai et al., 2017; Enge, Nylund, & Pavia, 2013; Orrock, dent, which adds further complexity and hampers predictability Dutra, Marquis, & Barber, 2015) and pathogens (Borer, Hosseini, of net indirect interaction strength at the community level. For Seabloom, & Dobson, 2007; Power & Mitchell, 2004), where exotic example, different groups of interaction partners (i.e. herbivores and native antagonists can spill over and spill back, respectively, to and pollinators) may indirectly influence one another via shared depress the growth of neighbouring native plants (Allen, Meyerson, host plants (Franzini, Azcon, Latanze-Mendes, & Aroca, 2010; Flick, & Cronin, 2018). Conversely, indirect interactions can link Koricheva, Gange, & Jones, 2009), and the strength of indirect species via shared mutualists, leading to several potential outcomes interactions that they mediate between plants may also be cor- (Dickie, Bufford, et al., 2017). For example, plants indirectly interact related (Fontaine & Thébault, 2015; Sauve, Thébault, Pocock, & with one another via shared fungal mutualists, generating both pos- Fontaine, 2016) because different organisms can respond similarly itive and negative impacts on plant performance via common myce- to plant cues (Theis, 2006). Thus, investigating pairwise interac- lial networks and altered fungal inoculum potential (Horton, 2015; tions in isolation, or even single guilds of interactions, risks over- Newman, 1988; van der Heijden & Horton, 2009). However, much of looking important indirect interactions among different community the literature has focused on the negative indirect impacts of invad- members. Indirect impacts of invasions have previously been stud- ers, with studies of indirect facilitation of native species via shared ied at the community level (e.g. Bumbeer, da Rocha, Bornatowski, mutualists largely limited to plant-pollinator systems (Bartomeus de Castro Robert, & Ainsworth, 2018; Feit et al., 2018; O'Dowd, et al., 2008; Charlebois & Sargent, 2017; but see Dickie et al., 2014), Green, & Lake, 2003), but largely through observation and mod- leaving it unclear whether indirect impacts of invasive species on elling rather than manipulative experiments. Investigating the native community members are systematically positive or negative. balance of these positive and negative indirect interactions could ALLEN et al. Journal of Ecolog y | 3 improve understanding and prediction of invasion success and im- Potter, Kriticos, Wait, & Leriche, 2009). Broom has high bio- pacts, yet no studies to date have compared the relative strength mass and density combined with diverse species interactions of these different types of interaction at a broader community (Lafay & Burdon, 2006; Memmott, Fowler, Paynter, Sheppard, & level (i.e. soil biota-plant-herbivore), or their impact across multiple Syrett, 2000), and hence strong potential for indirect impacts. To native and exotic taxa. measure the indirect impacts of broom via shared soil fungi and Here, we experimentally examine the indirect impacts of the arthropod herbivores, we used 21 plant species from the same global grassland invader Scotch broom Cytisus scoparius (L.) Link family as broom: Fabaceae (legumes; Table 1). We focused on (Fabaceae) (hereafter broom), on the survival and performance of 10 legumes to minimize phylogenetic distance from broom and thus native and 11 exotic legume species via shared soil fungi and herbi- improve the likelihood of them sharing some interaction partners, vores. By burying potted plants of each species either adjacent to or and our ability to detect indirect impacts. Many of the experimen- 50 m away from a large-scale broom invasion and manipulating the tal species share similar habitat or co-occur with broom in New presence of indirect pathways via exclusion of soil fungi and arthro- Zealand; however, co-occurrence was not a requirement for spe- pod herbivores, we tested the following predictions: (a) Broom im- cies selection, with species instead chosen to maximize phyloge- pacts the survival and growth of other plant species through indirect netic diversity within Fabaceae and based on availability of seed interactions mediated by mutualists and enemies; (b) Exotic plant or cuttings. species experience stronger direct and indirect impacts of broom than natives; and (c) Plants that experience strong indirect impacts of broom mediated by herbivores also experience strong indirect im- 2.3 | Experimental setup and design pacts mediated by soil fungi. There were four treatments in a fully crossed 2 × 2 × 2 × 2 fac- torial design with five replicate blocks and 21 plant species (block 2 | MATERIALS AND METHODS and species were random effects), resulting in a total of 840 potted plants. The first treatment was plant provenance: the species 2.1 | Study location used for the experiment comprized 10 native and 11 exotic species

The experiment was conducted at Brooksdale Station, Canterbury, New Zealand (−43.3067 N, 171.7673 W, elevation = 647 m). The TABLE 1 List of native and exotic legume species used in the field experiment site features a large (~0.5 km2) broom invasion bordering a montane grassland that is uninvaded by broom (Supporting Information S1, Plant species Common name Provenance

Figure S1). The uninvaded grassland community (hereafter, we use Carmichaelia appressa Prostrate broom Native the term uninvaded to refer to the community where broom is not Carmichaelia australis Common native Native present as an invader) was dominated by exotic Agrostis capillaris, broom Festuca rubra, Hieracium pilosella and diverse but sparse native sub- Carmichaelia stevensonii Weeping broom Native shrubs (Supporting Information S2, Table S1). This uninvaded com- Carmichaelia williamsii William's broom Native munity has a history of being grazed by livestock for several weeks Clianthus maximus Kākābeak Native in spring each year (but not during our experiment) and, importantly, Clianthus puniceus Kākābeak Native any broom seedlings are controlled by applying herbicide spray to Sophora longicarinata Limestone kōwhai Native individual plants every 2–3 years. This manual control of broom Sophora microphylla Small-leaved kōwhai Native means that the invasion front was human-controlled, rather than being driven by any hidden environmental gradients. In the broom- Sophora prostrata Prostrate kōwhai Native invaded community, broom was not managed and dominated a Sophora tetraptera Large-leaved kōwhai Native similar background community of A. capillaris, F. rubra and Dactylis Cytisus proliferus Tree lucerne Exotic glomerata, but few other plant species (Supporting Information S2, Cytisus multiflorus Portuguese broom Exotic Table S1, Figure S2). Cytisus scoparius Scotch broom Exotic Genista monspessulana Montpellier broom Exotic Lupinus arboreus Tree lupin Exotic 2.2 | Study species Lupinus polyphyllus Russell lupin Exotic Medicago sativa Lucerne Exotic Scotch broom is a perennial, nitrogen-fixing shrub that grows in Spartium junceum Spanish broom Exotic open and disturbed sites across a wide range of soil types. It is na- Trifolium pratense Red clover Exotic tive to western and central Europe but is invasive on all other con- Trifolium repens White clover Exotic tinents except Antarctica, with impacts on forestry, agriculture Ulex europaeus Gorse Exotic and native communities (Jarvis, Fowler, Paynter, & Syrett, 2006; 4 | Journal of Ecology ALLEN et al.

(including broom; Table 1; Figure 1). The second treatment was de- enabled comparison of the strength of herbivore or fungal interac- signed to manipulate fungal growth between pots and the surround- tions (i.e. the effect of the above three treatments) between invaded ing soil: pots were fitted with 10 cm × 6 cm nylon mesh windows of and uninvaded communities, which we interpreted as the indirect differing porosity (1 vs. 38 µm) to prevent or allow fungal hyphae impact of broom (i.e. apparent competition or indirect facilitation) growth, respectively (Figure 1) (Johnson, Leake, & Read, 2001; (see Box 1 for further interpretation details). Pots in the invaded Teste, Karst, Jones, Simard, & Durall, 2006). The third treatment community were situated on the edge of the broom invasion to min- was designed to exclude or allow impacts of arthropod herbivores: imize any shading effects that were not present in the uninvaded plants were sprayed with pyrethrum pesticide (Yates, Christchurch, grassland. A randomized blocked design was used to account for any New Zealand) or a water control. The fourth treatment was plant possible biotic or abiotic gradients along the transects, although we location: pots were buried adjacent to the broom invasion or 50 m observed no systematic changes in the dominant plant community in away in the uninvaded grassland (Figure 1; Figure S1). This treatment preliminary surveys (Supporting Information S2).

FIGURE 1 Summary of experimental design, illustrating the fully crossed (A) plant provenance treatment (native or exotic plant species); (B) spray treatment (pyrethrum pesticide spray to exclude arthropod herbivores or water control); (C) mesh size treatment (fungal hyphae can grow in and out of pots with 38 µm mesh, whereas 1 µm mesh blocks their growth); and (D) pot location treatment (adjacent to or 50 m away from Scotch broom Cytisus scoparius)

BOX 1 Interpretation of treatment main effects and interactions in mixed model analyses Main effects of location, mesh size or spray treatment with no interactions were interpreted as direct impacts of broom, soil fungi and arthropod herbivores, respectively. However, the location treatment main effect could also be driven by unmeasured indirect impacts of broom, potentially mediated by other taxa that we did not manipulate or quantify such as fungal endophytes or changes in the surrounding plant community. We interpreted the mesh size treatment as a fungal effect (meaning that fungi from outside the pot colonized the soil and experimental plant, fungi from inside the pot grew into the surrounding environment, or both), yet there are other possible interpretations that cannot be entirely ruled out, such as variable water drainage or movement of nutrients or other resources across the different mesh sizes. However, we may expect these abiotic impacts to be consistent in the invaded and uninvaded community and between native and exotic species, which was not supported by the results. Critically, we interpreted pairwise interactions of plant location with exclusion treatments (i.e. mesh or spray effects differ between the invaded and uninvaded community) as indirect impacts of broom, an approach that has been previously used to assess indirect impacts of invasive plants (Bhattarai et al., 2017). Similarly, because hare browsing, arthropod herbivore damage and rhizobia nodulation variables were not treatments themselves, we interpreted main effects of plant location for these variables as an indirect impact of broom mediated by changes in the density or behaviour of these taxa. We contend that broom is the major difference between the invaded and uninvaded plant communities (Supporting Information S2) and thus the main contributor to the differences in interactions observed between plants in the two locations. Effects involving plant provenance were interpreted as differences in inherent growth (main effect of provenance for plant performance variables), direct interactions (provenance × location/mesh/spray treatments) and indirect interactions (location × provenance × mesh/spray treatments) between native and exotic plants. Finally, interactions between the mesh and spray treatments were interpreted as host plant-mediated indirect interactions between soil fungi and arthropod herbivores. ALLEN et al. Journal of Ecolog y | 5

Plants were grown from seed (obtained from field populations, of belowground biomass and rhizobia nodulation (see below and New Zealand Tree Seeds, and Prebble Seeds, Christchurch, New Supporting Information S3 and S4 for details). Zealand), except Carmichaelia appressa and C. australis, which were propagated from stem cuttings treated with Clonex root hormone gel (Yates, Auckland, New Zealand). Following germination, seed- 2.5 | Data analysis lings were transplanted into pots on 16–17 November 2017 and maintained under regularly watered greenhouse conditions for We used model selection based on the Akaike Information Criterion 4 weeks. Pots were 1.5 L polyethylene terephthalate bottles (Alto, corrected for small sample size (AICc) to assess strength of direct Christchurch, New Zealand) with tops removed and a 2.5 cm layer of and indirect interactions in this montane grassland community, se- 5–10 mm diameter pebbles (Intelligro, Christchurch, New Zealand) lecting the best-fitting mixed-effect models from a set of candidate was added to the bottom of the pots to improve drainage. Soil was models for each response variable (Burnham & Anderson, 2010). mixed 50:50 with sand and was collected from the top 30 cm of the The full model for all response variables included plant provenance, uninvaded grassland at the field site to represent live soil naïve to pot location, mesh size, spray treatment and all possible interac- invasion and without any nitrogen increase associated with broom tions as fixed effects (15 total variables). Hare browsing was in- (Broadbent et al., 2017). Potted plants were established in the field cluded as a fixed covariate for analyses of belowground biomass between 13 and 23 December 2017, by fitting pots into 20 cm deep and nodulation. Random intercepts were included in all models holes drilled with a 10 cm diameter earth auger, ~1 m apart in a tran- for plant species and experimental block, to account for among- sect (Figure S1). Plant survival was assessed after two weeks and species variation and possible environmental gradients, respec- dead seedlings replaced. Seedlings were watered twice (January– tively. Candidate models were based on subsets of the full model, February 2018) or once (March–April) per week and weeded as using all possible combinations of the explanatory variables, but necessary. with the restrictions that main effects must also be present in models containing interactions, and that random effects were retained in every model combination to account for structural aspects of 2.4 | Data collection our design. We ranked candidate models from lowest to highest

AICc and models with ΔAICc (=AICci − AICcmin) of two or less were To assess initial growth, the height of all plants was recorded six deemed to have substantial support (Burnham & Anderson, 2010). weeks after transplanting into the field (5 February 2018), and was Each influential variable was then subjected to post-hoc Tukey measured again at harvest. Surviving plants (n = 570) were har- tests (Bonferroni corrected where appropriate). Different model vested between 16 and 30 April 2018. Roots were washed, and error distributions were used depending on the response variable, above- and belowground biomass separated into paper bags and and some variables were transformed to meet model assumptions dried for 72 hr at 55°C before being weighed. Using ImageJ software (Supporting Information S3, Table S2 and Supporting Information S4 (Rasband, 2018), we quantified cumulative arthropod herbivore for further analysis details). Although some plant performance vari- percent leaf damage from photographs of all leaves from each plant ables were correlated (Supporting Information S5, Table S3), we (excluding plants browsed by hares), estimating the undamaged report results for all because they represent different components leaf area by extrapolation. Finally, the number of root nodules was of plant growth that are of interest to researchers and that were counted for each plant as a proxy for plant–rhizobia interactions and expected to vary in their responses to different treatments. The standardized by root dry mass (nodules per gram of root dry mass) to total number of plants included in each analysis also varied depend- remove confounding effects of the experiment treatments on plant ing on the response variable being investigated, number of plants size, which may be correlated with the number of nodules. alive and the presence of hare browsing (Supporting Information S3, A few weeks into the experiment, we began finding plants that Table S2). Nodulation was assessed using a two-stage model, where were browsed down to near the base of the stem. Based on ob- we first examined treatments that were influential to the presence servations at the study site and the characteristic damage, we de- or absence of nodules, and then the number of nodules per gram of termined that European hares Lepus europaeus were responsible. root biomass for plants with nodules. To make effects comparable Rather than attempting to control the hares, we used their herbivory across different response variables, we calculated Cohen's d effect as an opportunity to separately assess the indirect impact of broom size (= model coefficient/standard deviation) for contrasts that were mediated by hares. Because it was impossible to determine how significant in post-hoc Tukey tests. The focus of the main text is on much plant tissue had been browsed, hare herbivory was quanti- comparing the size of these effects, but in the interest of transpar- fied as present or absent. Hare-browsed plants that still had healthy ency we present the full results and interpretation of the model se- roots or live aboveground tissue were not considered dead in the lection and post-hoc analyses in Supporting Information S6 and S7. survival assessment. Because hare herbivory was not a controlled Furthermore, to examine whether plant performance was influenced treatment, we excluded hare-browsed plants from analyses of plant by our observed direct interactions (i.e. hare herbivory, arthropod height, aboveground biomass, total biomass and arthropod herbi- damage and nodulation), we ran separate linear mixed models using vore damage, and included hare herbivory as a covariate for analyses total biomass as the response variable, the observed interaction as 6 | Journal of Ecology ALLEN et al. the explanatory variable, and plant species and experimental block Broom as random effects. Arthropod herbivores Hares Finally, we considered but elected not to use structural equation modelling to analyse the data, because incomplete observations for some response variables (e.g. arthropod herbivory, harvest height, aboveground biomass and total biomass could not be assessed for 6% leaf tissue damag hare-browsed plants) meant that a comprehensive SEM containing all response variables was not possible, and any reduced analysis would result in a significant loss of data and inference power for hy- e

tal biomass (0.15) To

Hare browsing (0.17) browsing Hare Belowground biomass (0.10) biomass Belowground pothesis tests of all pathways. (0.32) height Harvest Native plants Native and exotic plants Exotic plants We also extracted species-level direct and indirect impacts of broom from the analyses to test whether (a) direct or indirect impacts of broom were stronger; and (b) indirect impacts mediated by herbivores and soil fungi were correlated. To address these questions, we estimated species-level coefficients by including random slopes for each experimental plant species in best-fitting models where direct

(i.e. location main effects for plant performance variables) or indi- Belowground biomass (0.13) rect (i.e. a location × mesh or location × spray interaction for plant Aboveground biomass (0.15) performance variables or a location main effect for observed arthro- tal biomass (0.29) To pod herbivore damage, hare herbivory and nodulation) impacts of Height at 6 weeks (0.18) broom were identified. Using these species-level coefficients, we first tested for differences in the strength of direct and indirect in- Negative direct Belowground biomass (0.27) effect teractions using a linear mixed model, with interaction type (direct Harvest height (0.29) Positive direct vs. indirect) as the sole fixed effect and plant species and dependent effect SiSoillf fungii variable identity (e.g. total biomass, survival, arthropod herbivory) (root tip and hyphae) as random effects in the model. Second, we used linear regression FIGURE 2 Summary of all measured direct effects of Scotch to ask whether plants that experienced strong indirect impacts of broom Cytisus scoparius, herbivores, soil fungi and rhizobia on broom mediated by herbivores also experienced strong indirect im- native and exotic plant species when grown adjacent to broom. pacts mediated by soil fungi. We also tested whether indirect im- Arrows indicate influential main effects retained in best-fitting pacts mediated by arthropod herbivores and hares were correlated. models and their colour reflects positive (blue) or negative Finally, we also tested whether phylogenetic relatedness predicted (red) direct impacts on plants from the respective interaction the strength of indirect interactions (i.e. species-level random effect partner. Arrow width represents Cohen's d effect sizes (i.e. direct interaction strength, quantified in parentheses) calculated from coefficients). However, because of the limited breadth (i.e. within a regression coefficients for the main effect of location (broom), single family, Fabaceae) and resolution (i.e. only to genus level) of our mesh size (soil fungi) and plant provenance (hare browsing and phylogeny, we elected to present these results only in Supporting rhizobia nodule presence). The dotted arrow from arthropod Information S8. Absolute values of interaction strength (i.e. random herbivores indicates that leaf damage was observed but that the slope coefficients) were used throughout these analyses because we spray treatment did not influence plant performance, and thus a Cohen's d effect size could not be calculated were interested in the effect magnitude rather than direction, and indirect interaction strength could be both positively and negatively influenced depending on the context (e.g. beneficial indirect facilita- model selection and post-hoc analyses, including test statistics and tion vs. harmful mutualist competition). All analyses were performed p-values, are reported in Supporting Information S6 and S7. There in R 3.6.0. (R Development Core Team, 2019) using the lme4 (Bates was a main effect of location (with no interactions) for several vari- et al., 2019), MuMIn (Bartoń, 2019), and emmeans (Lenth, Singmann, ables, suggesting that broom had a positive direct impact (i.e. likely Love, Buerkner, & Herve, 2019) packages. not via above- or belowground interaction partners, although this cannot be entirely excluded; see Box 1) on total biomass (35% increase), belowground biomass (20% increase) and height at harvest 3 | RESULTS (40% increase) of both native and exotic experimental plant species (Cohen's d = 0.15, 0.10 and 0.32, respectively; Figure 2; Figure S5). 3.1 | Most direct interactions were stronger for There was an interaction between mesh size and plant prove- exotic than native plants nance, indicating that soil fungi had a direct positive impact on height after 6 weeks (54% increase), harvest height (53%), total Plant performance was influenced by many direct (Figure 2) and in- biomass (112%) and aboveground biomass (111%) for exotic but not direct interactions (Figure 3). Detailed results and interpretation of native plants (Cohen's d = 0.18, 0.29, 0.29 and 0.15, respectively; ALLEN et al. Journal of Ecolog y | 7

Arthropod herbivores Hares

% leaf tissue damage (0.21)

Hare browsing (–0.25) Hare browsing (–0.10) browsing Hare Native plants Native and exotic plants Exotic plants

(0.09)

(0.15)

)

tal biomass (0.28) biomass tal Survival (0.14) Survival

Aboveground biomass (0.23) biomass Aboveground weeks (0.23

Harvest height (0.29) height Harvest To Nodule presence (0.12) presence Nodule Negative indirect

effect of broom Height at 6 Positive indirect

effect of broom (–0.24) Other negative indirect effect Soil fungi Rhizobia (root tip and hyphae)

FIGURE 3 Indirect impacts of broom on native and exotic plant species via their shared interaction partners, as determined by the difference in strength of the interaction partner effect between locations (invaded, uninvaded) in the field experiment. Arrows indicate influential effects retained in best-fitting models. Arrow colour reflects whether broom has a positive (blue) or negative (red) indirect impact on the plant, mediated via the corresponding interaction partner. Purple arrows represent negative consequences that have been amplified via indirect impacts of soil fungi or rhizobia on other interaction partners, mediated by the experimental plant. Arrow width represents Cohen's d effect sizes (i.e. indirect interaction strength, quantified in parentheses) FIGURE 4 Impact of fine and coarse pot mesh on height at calculated from regression coefficients and shows the magnitude of harvest (estimated marginal mean ± 95% CI) (A), total biomass (B), the effect of broom (i.e. location) on the interaction aboveground biomass (C) and belowground biomass (D) for native and exotic plant species and on the probability of plant survival (E), total biomass (F), aboveground biomass (G) and leaf tissue damaged by arthropod herbivores (H) in habitat invaded or uninvaded by Figure 4A–C, Figures S4 and S6). Soil fungi also increased below- Scotch broom Cytisus scoparius. Different lowercase letters indicate ground biomass of native plants, but the effect was twice as strong significant differences among means for each variable in post-hoc for exotics (41% and 98% increase, respectively; Cohen's d = 0.13 Tukey tests (p < 0.05). The interactions of mesh size with location and 0.27, Figure 4D). Furthermore, the number of rhizobia nodules and plant provenance were identified as influential using mixed- per gram of root biomass was 70% higher for exotic than native effects model selection (Table S4) plants, although this effect was not strong enough to be significant (Figure S11C). nodules than undamaged plants (Cohen's d = 0.09; Figure S11B) and Aboveground, 44% of plants were browsed by hares. Broom in- 38% less biomass (Figure S7). In contrast to hare browsing, arthropod vasion decreased the probability of hare herbivory by 39% and 81% herbivory was low, averaging just 6.3 ± 1.1% of leaf area damaged, for exotic and native species, respectively, resulting in three times did not differ between native and exotic plants, and was unrelated higher hare browsing of exotic than native plants in the invaded to total biomass. Few herbivores or evidence of their feeding was community (Cohen's d = 0.17; Figure 5). The pyrethrum spray treat- observed other than several species of Orthoptera, minor gastropod ment increased hare herbivory of plants in the uninvaded habitat damage, one leafroller caterpillar (Lepidoptera: Tortricidae), and the (46% higher), native plants (105% higher) and those grown in pots distinctive leaf-notching damage of clover root weevil (Sitona obsole- with fine mesh (87% higher; Cohen's d = 0.16, 0.14 and 0.16, respec- tus, Coleoptera: Curculionidae) on foliage of three Trifolium individ- tively; Figure S9). Plants browsed by hares on average had 15% more uals. The pyrethrum spray treatment had no direct impact on any of 8 | Journal of Ecology ALLEN et al.

(Cohen's d = 0.12; Figure S10), but this difference was only statistically significant when adjacent to broom or sprayed with the water control. Nodule abundance was also 40% higher when soil fungi were excluded via the fine mesh treatment (Cohen's d = −0.24; Figure S11A). Overall, the indirect impacts of broom were over six times stronger than the direct impacts that we measured at the spe-

cies level (F1,206 = 66.83, p < 0.001). Finally, we found no correlation between the indirect impacts of broom on each plant species that were mediated by herbivores and soil fungi (Supporting Information S8, Table S5).

4 | DISCUSSION

Our field experiment demonstrates how invader impacts occur through many direct and indirect pathways above- and belowground, with stronger positive and negative impacts on other exotic species than natives (summarized in Figures 2 and 3). We found that despite FIGURE 5 The probability of European hare Lepus europaeus browsing (estimated marginal mean ± 95% CI) for native and exotic increasing arthropod herbivory, broom caused a net increase in sur- plants in communities invaded and uninvaded by Scotch broom vival and growth of other plant species, both directly, likely through Cytisus scoparius. Different lowercase letters indicate significant shelter from abiotic stress, and indirectly, via beneficial soil fungi and differences among means in post-hoc Tukey tests (p < 0.05). The release from hare browsing (Figures 4 and 5). Soil fungi associated with interaction between plant provenance and location was identified broom also had negative indirect impacts through increased arthropod as influential using mixed-effect model selection (Table S4) herbivory (Figure 4H) and decreased rhizobia nodulation (Figure S11). Overall, the direct and indirect impacts we observed were mostly pos- the plant performance variables, but reduced herbivore damage by itive and favoured exotic plants belowground (via soil fungi and rhizo- 26%, although this effect was non-significant in post-hoc analyses bia) and native plants aboveground (via herbivores). Moreover, the (Figure S8). indirect impacts of broom that we measured were six times stronger than direct impacts, highlighting the importance of both direct and indirect interactions in driving invasion impacts. Finally, we found little 3.2 | Indirect interactions were mostly positive and evidence that the strength of indirect interactions can be predicted by favoured exotic plants belowground and native plants the strength of other indirect interactions (Table S5), indicating that a aboveground more nuanced approach may be required for future studies. Survival and growth of other plant species were higher in the Aboveground, broom invasion had a positive indirect impact on invaded community, supporting the concept of ‘nurse plants’ that other plants through decreased hare herbivory relative to the un- facilitates the growth of other species (e.g. Burrows, Cieraad, & invaded community, which was stronger for native (81% decrease) Head, 2015; Pugnaire et al., 1996). We extend this concept by iden- than exotic plants (39%) (Cohen's d = −0.25 and −0.10, respectively; tifying potential mechanisms: (a) direct impacts (Figure 2), possi- Figures 3 and 5). Conversely, broom invasion increased arthropod bly through shelter from abiotic stress (Carter, Slesak, Harrington, herbivore damage by eightfold relative to the uninvaded community, Peter, & D'Amato, 2019), and (b) indirect impacts, mediated by ben- but only for plants allowed to interact with external fungi via coarse eficial soil fungi, rhizobia and escape from hare browsing (Figure 3). mesh (Cohen's d = 0.21; Figures 3 and 4H). The indirect impact of The lower hare browsing in the invaded community (Figure 5) was broom on plant height after 6 weeks via belowground interaction likely behaviourally mediated, due to decreased need for predator partners was positive (54% increase) for exotic plants (Cohen's vigilance and thereby better browsing in open areas (Marboutin & d = 0.23; Figure S4) but no effect was observed for native plants. Aebischer, 1996), or the denser broom and grass vegetation limiting However, the indirect impact of broom via belowground interaction hares' ability to visually locate plants. Our result showing facilitative partners on survival, total biomass, aboveground biomass and height effects of broom differs from studies that have demonstrated neg- at harvest was positive for both native and exotic plants (Cohen's ative impacts of invasive plants on native plants through apparent d = 0.14, 0.28, 0.23 and 0.29 respectively; Figure 4E–G; Figure S6). competition mediated by both mammalian and arthropod herbivores In addition, soil fungi associated with broom increased arthropod (e.g. Bhattarai et al., 2017; Enge et al., 2013; Orrock et al., 2015). herbivore damage by 133% (Cohen's d = 0.15; Figure 4H). Finally, the However, broom also increased arthropod herbivory eightfold probability of rhizobia nodule presence on roots was 0.99 ± 0.0001 (Figures 3 and 4H), although total leaf damage remained low and was (mean ± SE) for exotic plants compared to 0.78 ± 0.001 for natives not correlated with plant fitness. These contrasting results could be ALLEN et al. Journal of Ecolog y | 9 because arthropod herbivory measured cumulative damage from the outcome, finding that herbaceous grassland species tended to as- entire community, including both generalists and specialists, whereas sociate with more exotic plant species in their introduced range hare herbivory was from one generalist that browsed all 21 species. and with more native plant species in their native range, a pattern Regardless, these differences highlight how investigating single in- that could result from indirect impacts of invasive plants that favour teractions and native-invasive species pairs precludes making broad other exotic species over natives. conclusions about invader impacts. Here, we present invasive species Aboveground, our findings of greater hare browsing on exotic impacts in a broader community context by investigating several in- than native plants and no difference in arthropod herbivore dam- teractions involving 21 species of native and exotic plants, finding age both conflict with the enemy release hypothesis, which posits that broom impacts were generally consistent within plant prove- that exotic species are successful because they suffer less dam- nances rather than the result of idiosyncratic species pairing. age from enemies relative to native species (Elton, 1958; Keane While aboveground indirect impacts of broom may have been & Crawley, 2002). However, our results are consistent with evi- mediated by hare behaviour, the positive belowground indirect im- dence suggesting that exotic plants can suffer greater herbivory pacts on neighbours (Figures 3 and 4) were likely a consequence than natives (Parker, Burkepile, & Hay, 2006; Parker & Hay, 2005; of broom increasing the quantity and/or quality of beneficial soil Waller et al., 2020), supporting the biotic resistance hypothesis biota (e.g. mycorrhizal fungi and rhizobia bacteria; Horton, 2015; (Elton, 1958). Interestingly, broom, hares and nine of the eleven ex- Newman, 1988). Although the outcome of mycorrhiza-mediated in- otic plant species were all introduced from Europe to New Zealand, teractions between mature plants and seedlings is most frequently where they also regularly co-occur, suggesting that coevolutionary facilitative (48% of studies, van der Heijden & Horton, 2009), neg- history in both their native and introduced ranges could contrib- ative interactions such as mycorrhiza-mediated growth depres- ute to the stronger direct and indirect impacts that we observed sion in seedlings competing with a stronger host (Waller, Callaway, for the exotic than native plant species. Regardless of herbivore Klironomos, Ortega, & Maron, 2016) can also occur (25% of stud- provenance, if exotic plants accumulate high densities of general- ies), as well as no impact whatsoever (27%). We also cannot rule ist herbivores and are more tolerant of this herbivory (Ashton & out the possibility that fungal-mediated impacts could occur via Lerdau, 2008), this may suggest high potential for strong indirect direct and indirect impacts of broom on other plants in the com- impacts on the surrounding community. Moreover, as the density munity and their fungal interaction partners (i.e. the broader plant- of the invader and its interaction partners increases, we may expect mycorrhizal network), which then fed back onto our experimen- both positive and negative indirect impacts to intensify as an inva- tal seedlings. Furthermore, soil fungi also had their own negative sion progresses. Therefore, we suggest that management practices indirect impacts via increasing arthropod herbivory in the broom- targeting shared interaction partners as well as the focal invader invaded community (Figures 3 and 4H), likely driven by increased may improve mitigation of negative invasion impacts or promotion plant nutritional value or mycorrhizal suppression of plant defences of positive impacts. For example, in Otago, New Zealand, the inva- (Koricheva et al., 2009). Plants that interacted with soil fungi also sive venomous redback spider Latrodectus hasseltii is facilitated by had lower presence of rhizobia nodules (Figure 3; Figure S11), possi- burrows of another invader, the European rabbit Oryctolagus cunic- bly due to competition between rhizobia and mycorrhiza for limited ulus, and preys upon the critically endangered endemic Cromwell carbohydrates of the young seedlings (Franzini et al., 2010). chafer beetle Prodontria lewisii. By filling in rabbit burrows, spider Despite the complexity of interactions described above, some density and capture of the chafer beetle were decreased to zero consistent patterns emerged. For example, the direct and indirect im- (Spencer, van Heezik, Seddon, & Barratt, 2017). pacts we measured were mostly positive and favoured exotic plants Finally, in contrast to our expectations, we found that the belowground and native plants aboveground, with a net advantage strength of indirect impacts mediated by other interaction partners to exotic plants. Belowground, exotic plants experienced stronger (i.e. soil fungi vs. hares) was generally a poor predictor of indirect positive interactions with soil fungi and rhizobia, potentially through impacts of broom (Table S5). This result was somewhat surpris- greater generality and promiscuity with interaction partners (Moora ing in light of previous studies that have demonstrated that spe- et al., 2011) or being better hosts (Lekberg, Gibbons, Rosendahl, & cies interactions are sometimes correlated between mutualistic Ramsey, 2013). The stronger indirect facilitation of exotic plants by and antagonistic partners (Sauve et al., 2016). However, it may be broom via soil fungi and rhizobia could represent an indirect mech- that functional trait data specific to the species interactions inves- anism of invasional meltdown (Simberloff & Von Holle, 1999), po- tigated (e.g. plant defences, specific root length and root diameter) tentially leading to ecosystems that are increasingly dominated by (Carmona, Lajeunesse, & Johnson, 2011; Eissenstat, Kucharski, exotic species. A meta-analysis by Kuebbing and Nuñez (2016) found Zadworny, Adams, & Koide, 2015) would better predict indirect in- similar evidence for competitive interactions, showing that negative teraction strength among species, especially when combined with competitive impacts of exotic plants on native plants are two times phylogenetically informed quantitative interaction network analyses stronger than their impacts on other exotic plants, further suggest- (e.g. Frost et al., 2016; Pearse & Altermatt, 2013; Tack, Gripenberg, ing that indirect interactions among multiple native and non-native & Roslin, 2011). species could favour exotic dominance of plant communities. A re- There are several limitations of our study that we would like to cent global analysis by Stotz et al. (2020) supports this hypothesized acknowledge. First, although we have simultaneously quantified 10 | Journal of Ecology ALLEN et al. multiple direct and indirect impacts of broom on 21 native and ex- Commission and the Royal Society of New Zealand Hutton Fund. otic species, these were the impacts of a single invasive species at The authors declare no conflict of interest. one site and over a single growing season. Second, all of the experimental species were legumes in order to minimize phylogenetic AUTHORS' CONTRIBUTIONS distance from broom and maximize their likelihood of sharing inter- I.A.D. and J.M.T. obtained funding; W.J.A., R.W., J.M.T., B.I.P.B. and action partners, thereby limiting our ability to extend our findings to I.A.D. designed the experiment; W.J.A., R.W., M.-R.S. and L.P.W. set other plant families and meaning that inference outside of the family up the experiment and collected data; W.J.A. conducted analyses Fabaceae should be treated with caution. Moreover, this selection of and wrote the first draft of the manuscript, and all authors contrib- closely related species could potentially generate bias towards the uted substantially to revisions. detection of strong indirect effects based on phylogenetic conservatism of species interactions (Chagnon, Bradley, & Klironomos, 2015; DATA AVAILABILITY STATEMENT Peralta, 2016), although we found no evidence of this phyloge- Data are available from the Dryad Digital Repository https://doi. netic relationship within Fabaceae (Supporting Information S8, org/10.5061/dryad.2v6wwpzjg (Allen et al., 2020). Figure S12). Thus, although these caveats may limit the inference space for invasions in general, we believe that our findings are at the ORCID forefront of scaling up experimental invasion ecology to the commu- Warwick J. Allen https://orcid.org/0000-0002-1859-1668 nity level and present a framework for predictions that can be tested Barbara I. P. Barratt https://orcid.org/0000-0002-8424-8729 in other systems, while also providing critical local information on Lauren P. Waller https://orcid.org/0000-0002-7110-6027 broom invasion and impacts in New Zealand. Ian A. Dickie https://orcid.org/0000-0002-2740-2128 We conclude that invader impacts occur through the balance of multiple direct and indirect pathways involving soil fungi, rhizobia REFERENCES and mammalian and arthropod herbivores. In our system, broom had Allen, W. J., Meyerson, L. A., Flick, A. J., & Cronin, J. T. (2018). Intraspecific contrasting impacts on native and exotic plant species above and variation in indirect plant–soil feedbacks influences a wetland plant invasion. Ecology, 99, 1430–1440. https://doi.org/10.1002/ below ground, but with a net advantage to exotics, suggesting that ecy.2344 direct and indirect interactions may contribute to invasional melt- Allen, W. J., Wainer, R., Shadbolt, M.-W., Tylianakis, J. M., Barratt, B. I. down of exotic plants. Furthermore, although the net outcome for P., Waller, L. P., & Dickie, I. A. (2020). Data from: Community-level species interacting with broom depends on both direct and indirect direct and indirect impacts of an invasive plant favour exotic over native species. Dryad Digital Repository, https://doi.org/10.5061/dryad. impacts, we found that indirect interactions acted through multi- 2 v 6 w w p z j g ple mediating species and were over six times as strong as direct Ashton, I. W., & Lerdau, M. T. (2008). Tolerance to herbivory, and not re- impacts, indicating that management practices should consider tar- sistance, may explain differential success of invasive, naturalized, and geting shared interaction partners as well as the focal invader. Our native North American temperate vines. Diversity and Distributions, results integrate multiple invasion hypotheses that invoke species 14, 169–178. https://doi.org/10.1111/j.1472-4642.2007.00425.x Bartomeus, I., Vilà, M., & Santamaría, L. (2008). Contrasting effects of interactions, and highlight the complex and unpredictable role of in- invasive plants in plant-pollinator networks. Oecologia, 155, 761–770. direct interactions in driving invasion impacts on above and below- https://doi.org/10.1007/s00442-007-0946-1 ground communities as well as the importance of a community-level Bartoń, K. (2019). MuMIn: Multi-model inference. R package version approach to studying biological invasions. Through taking a more 1.43.6. Retrieved from http://CRAN.R-project.org/package=MuMIn Bates, D., Maechler, M., Bolker, B., Walker, S., Christensen, R. H. B., holistic future perspective we will improve our understanding of the Singmann, H., & Fox, J. (2019). lme4: Linear mixed-effects models role biotic interactions play in the success of invaders and their im- using ‘Eigen’ and S4. R package version 1.1-21. Retrieved from http:// pacts on natural communities. CRAN.R-project.org/package=lme4 Bhattarai, G. P., Meyerson, L. A., & Cronin, J. T. (2017). Geographic variation in apparent competition between native and invasive Phragmites ACKNOWLEDGEMENTS australis. Ecology, 98, 349–358. https://doi.org/10.1002/ecy.1646 The authors thank Brooksdale Station and Stu Gunn for field site Borer, E. T., Hosseini, P. R., Seabloom, E. W., & Dobson, A. P. (2007). access and to Hamish Maule for assisting with vegetation surveys Pathogen-induced reversal of native dominance in a grassland and plant identification. We are also grateful to Jim Allen, Neroli community. Proceedings of the National Academy of Sciences of the United States of America, 104, 5473–5478. https://doi.org/10.1073/ Allen, Thomas Carlin, David Conder, Daniel Dash, Brian Kwan, pnas.0608573104 Aimee McKinnon, Leona Meachen, Brent Richards, Georgia Steel Braga, R. R., Gómez-Aparicio, L., Heger, T., Vitule, J. R. S., & Jeschke, J. M. and Angela Wakelin, for field and lab assistance, and to the ABiNZE, (2018). Structuring evidence for invasional meltdown: Broad support Weed Wing, Stouffer, Tylianakis and Ecosystem Mycology research but with biases and gaps. Biological Invasions, 20, 923–936. https:// doi.org/10.1007/s10530-017-1582-2 groups for helpful discussions about the experiment. We also thank Broadbent, A. A. D., Orwin, K. H., Peltzer, D. A., Dickie, I. A., Mason, N. James Cronin, Ganesh Bhattarai and the anonymous reviewers W. H., Ostle, N. J., & Stevens, C. J. (2017). Invasive N-fixer impacts for their feedback on earlier versions of the manuscript. This pro- on litter decomposition driven by changes to soil properties not litter ject was supported by Centre of Research Excellence funding to quality. Ecosystems, 20, 1151–1163. https://doi.org/10.1007/s1002 the Bio-Protection Research Centre from the Tertiary Education 1-016-0099-3 ALLEN et al. Journal of Ecolog y | 11

Bumbeer, J., da Rocha, R. M., Bornatowski, H., de Castro Robert, M., & Holt, R. D. (1977). Predation, apparent competition, and the structure Ainsworth, C. (2018). Predicting impacts of lionfish (Pterois volitans) of prey communities. Theoretical Population Biology, 12, 197–229. invasion in a coastal ecosystem of southern Brazil. Biological Invasions, https://doi.org/10.1016/0040-5809(77)90042-9 20, 1257–1274. https://doi.org/10.1007/s10530-017-1625-8 Holt, R. D., & Bonsall, M. B. (2017). Apparent competition. Annual Review Burnham, K. P., & Anderson, D. R. (2010). Model selection and multimodel of Ecology, Evolution, and Systematics, 48, 447–471. https://doi. inference: A practical information-theoretic approach (2nd ed.). New org/10.1146/annur e v - e c o l s y s - 1 1 0 3 1 6 - 0 2 2 6 2 8 York, NY: Springer. Horton, T. R. (2015). Mycorrhizal networks. Dordrecht, The Netherlands: Burrows, L., Cieraad, E., & Head, N. (2015). Scotch broom facilitates in- Springer. digenous tree and shrub germination and establishment in dryland Jarvis, P. J., Fowler, S. V., Paynter, Q., & Syrett, P. (2006). Predicting the eco- New Zealand. New Zealand Journal of Ecology, 39, 61–70. nomic benefits and costs of introducing new biological control agents Carmona, D., Lajeunesse, M. J., & Johnson, M. T. J. (2011). Plant traits for Scotch broom Cytisus scoparius into New Zealand. Biological Control, that predict resistance to herbivores. Functional Ecology, 25, 358– 39, 135–146. https://doi.org/10.1016/j.biocontrol.2006.07.012 367. https://doi.org/10.1111/j.1365-2435.2010.01794.x Johnson, D., Leake, J. R., & Read, D. J. (2001). Novel in-growth core Carter, D. R., Slesak, R. A., Harrington, T. B., Peter, D. H., & D'Amato, A. system enables functional studies of grassland mycorrhizal my- W. (2019). Scotch broom (Cytisus scoparius) modifies microenviron- celial networks. New Phytologist, 152, 555–562. https://doi.org/ ment to promote nonnative plant communities. Biological Invasions, 10.1046/j.0028-646X.2001.00273.x 21, 1055–1073. https://doi.org/10.1007/s10530-018-1885-y Keane, R. M., & Crawley, M. J. (2002). Exotic plant invasions and the Chagnon, P.-L., Bradley, R. L., & Klironomos, J. N. (2015). Trait-based enemy release hypothesis. Trends in Ecology and Evolution, 17, 164– partner selection drives mycorrhizal network assembly. Oikos, 124, 170. https://doi.org/10.1016/S0169-5347(02)02499-0 1609–1616. https://doi.org/10.1111/oik.01987 Koricheva, J., Gange, A. C., & Jones, T. (2009). Effects of mycorrhizal Charlebois, J. A., & Sargent, R. D. (2017). No consistent pollinator- fungi on insect herbivores: A meta-analysis. Ecology, 90, 2088–2097. mediated impacts of alien plants on natives. Ecology Letters, 20, https://doi.org/10.1890/08-1555.1 1479–1490. https://doi.org/10.1111/ele.12831 Kuebbing, S. E., & Nuñez, M. A. (2016). Invasive non-native plants have Dickie, I. A., Bufford, J. L., Cobb, R. C., Desprez-Loustau, M.-L., Grelet, a greater effect on neighbouring natives than other non-natives. G., Hulme, P. E., … Williams, N. M. (2017). The emerging science of Nature Plants, 2, 16134. https://doi.org/10.1038/nplants.2016.134 linked plant–fungal invasions. New Phytologist, 215, 1314–1332. Lafay, B., & Burdon, J. J. (2006). Molecular diversity of rhizobia https://doi.org/10.1111/nph.14657 nodulating the invasive legume Cytisus scoparius in Australia. Dickie, I. A., Cooper, J. A., Bufford, J. L., Hulme, P. E., & Bates, S. T. Journal of Applied Microbiology, 100, 1228–1238. https://doi. (2017). Loss of functional diversity and network modularity in in- org/10.1111/j.1365-2672.2006.02902.x troduced plant–fungal symbioses. AoB Plants, 9, plw084. https://doi. Lekberg, Y., Gibbons, S. M., Rosendahl, S., & Ramsey, P. W. (2013). Severe org/10.1093/aobpla/plw084 plant invasions can increase mycorrhizal fungal abundance and di- Dickie, I. A., St John, M. G., Yeates, G. W., Morse, C. W., Bonner, K. I., versity. The ISME Journal, 7, 1424–1433. https://doi.org/10.1038/ Orwin, K., & Peltzer, D. A. (2014). Belowground legacies of Pinus ismej.2013.41 contorta invasion and removal result in multiple mechanisms of in- Lenth, R., Singmann, H., Love, J., Buerkner, P., & Herve, M. (2019). em- vasional meltdown. AoB Plants, 6, plu056. https://doi.org/10.1093/ means: Estimated marginal means, aka least-squares means. R package aobpla/plu056 version 1.3.5.1. Retrieved from http://CRAN.R-project.org/package= Eissenstat, D. M., Kucharski, J. M., Zadworny, M., Adams, T. S., & Koide, emmeans R. T. (2015). Linking root traits to nutrient foraging in arbuscular my- Marboutin, E., & Aebischer, N. J. (1996). Does harvesting arable crops in- corrhizal trees in a temperate forest. New Phytologist, 208, 114–124. fluence the behaviour of the European hare Lepus europaeus? Wildlife https://doi.org/10.1111/nph.13451 Biology, 2, 83–91. https://doi.org/10.2981/wlb.1996.036 Elton, C. S. (1958). The ecology of invasions by animals and plants. London, Memmott, J., Fowler, S. V., Paynter, Q., Sheppard, A. W., & Syrett, P. UK: Methuen. (2000). The invertebrate fauna on broom, Cytisus scoparius, in two Enge, S., Nylund, G. M., & Pavia, H. (2013). Native generalist herbivores native and two exotic habitats. Acta Oecologia, 21, 213–222. https:// promote invasion of a chemically defended seaweed via refuge- doi.org/10.1016/S1146 -609X(00)00124 - 7 mediated apparent competition. Ecology Letters, 497, 487–492. Moora, M., Berger, S., Davison, J., Öpik, M., Bommarco, R., Bruelheide, H., … https://doi.org/10.1111/ele.12072 Walther, G.-R. (2011). Alien plants associate with widespread generalist Feit, B., Gordon, C. E., Webb, J. K., Jessop, T. S., Laffan, S. W., arbuscular mycorrhizal fungal taxa: Evidence from a continental-scale Dempster, T., & Letnic, M. (2018). Invasive cane toads might initi- study using massively parallel 454 sequencing. Journal of Biogeography, ate cascades of direct and indirect effects in a terrestrial ecosystem. 38, 1305–1317. https://doi.org/10.1111/j.1365-2699.2011.02478.x Biological Invasions, 20, 1833–1847. https://doi.org/10.1007/s1053 Newman, E. I. (1988). Mycorrhizal links between plants: Their function- 0-018-1665-8 ing and ecological significance. Advances in Ecological Research, 18, Fontaine, C., & Thébault, E. (2015). Comparing the conservatism of eco- 243–270. https://doi.org/10.1016/S0065-2504(08)60182-8 logical interactions in plant–pollinator and plant–herbivore networks. O'Dowd, D. J., Green, P. T., & Lake, P. S. (2003). Invasional ‘meltdown’ Population Ecology, 57, 29–36. https://doi.org/10.1007/s10144- on an oceanic island. Ecology Letters, 6, 812–817. https://doi. 014-0473-y org/10.1046/j.1461-0248.2003.00512.x Franzini, V. I., Azcon, R., Latanze-Mendes, F., & Aroca, R. (2010). Orrock, J. L., Dutra, H. P., Marquis, R. J., & Barber, N. A. (2015). Apparent Interaction between Glomus species and Rhizobium strains af- competition and native consumers exacerbate the strong competi- fect the nutritional physiology of drought stressed legume hosts. tive effect of an exotic plant species. Ecology, 96, 1052–1061. https:// Journal of Plant Physiology, 167, 614–619. https://doi.org/10.1016/ doi.org/10.1890/14-0732.1 j.jplph.2009.11.010 Parker, J. D., Burkepile, D. E., & Hay, M. E. (2006). Opposing effects of Frost, C. M., Peralta, G., Rand, T. A., Didham, R. K., Varsani, A., & native and exotic herbivores on plant invasions. Science, 311, 1459– Tylianakis, J. M. (2016). Apparent competition drives community- 1461. https://doi.org/10.1126/science.1121407 wide parasitism rates and changes in host abundance across eco- Parker, J. D., & Hay, M. E. (2005). Biotic resistance to plant invasions? system boundaries. Nature Communications, 7, 12644. https://doi. Native herbivores prefer non-native plants. Ecology Letters, 8, 959– org/10.1038/ncomms12644 967. https://doi.org/10.1111/j.1461-0248.2005.00799.x 12 | Journal of Ecology ALLEN et al.

Pearse, I. S., & Altermatt, F. (2013). Predicting novel trophic interactions Teste, F. P., Karst, J., Jones, M. D., Simard, S. W., & Durall, D. M. (2006). in a non-native world. Ecology Letters, 16, 1088–1094. https://doi. Methods to control ectomycorrhizal colonization: Effectiveness of org/10.1111/ele.12143 chemical and physical barriers. Mycorrhiza, 17, 51–65. https://doi. Peralta, G. (2016). Merging evolutionary history into species inter- org/10.1007/s00572-006-0083-4 action networks. Functional Ecology, 30, 1917–1925. https://doi. Theis, N. (2006). Fragrance of Canada thistle (Cirsium arvense) attracts org/10.1111/1365-2435.12669 both floral herbivores and pollinators. Journal of Chemical Ecology, 32, Pimentel, D., Zuniga, R., & Morrison, D. (2005). Update on the environ- 917–927. https://doi.org/10.1007/s10886-006-9051-x mental and economic costs associated with alien-invasive species Tylianakis, J. M., Didham, R. K., Bascompte, J., & Wardle, D. A. (2008). in the United States. Ecological Economics, 52, 273–288. https://doi. Global change and species interactions in terrestrial ecosystems. org/10.1016/j.ecolecon.2004.10.002 Ecology Letters, 11, 1351–1363. https://doi.org/10.1111/j.1461-0248. Potter, K. J. B., Kriticos, D. J., Wait, M. S., & Leriche, A. (2009). The current 2008.01250.x and future potential distribution of Cytisus scoparius: A weed of pasto- van der Heijden, M. G. A., & Horton, T. R. (2009). Socialism in soil? The ral systems, natural ecosystems and plantation forestry. Weed Research, importance of mycorrhizal fungal networks for facilitation in nat- 49, 271–282. https://doi.org/10.1111/j.1365-3180.2009.00697.x ural ecosystems. Journal of Ecology, 97, 1139–1150. https://doi. Power, A. G., & Mitchell, C. E. (2004). Pathogen spillover in disease org/10.1111/j.1365-2745.2009.01570.x epidemics. The American Naturalist, 164, S79–S89. https://doi. van Kleunen, M., Weber, E., & Fischer, M. (2010). A meta-analysis of trait org/10.1086/424610 differences between invasive and non-invasive plant species. Ecology Pugnaire, F. I., Haase, P., Puigdefábregas, J., Cueto, M., Clark, S. C., Incoll, Letters, 13, 235–245. https://doi.org/10.1111/j.1461-0248.2009. L. D., & Puigdefabregas, J. (1996). Facilitation and succession under 01418.x the canopy of a leguminous shrub, Retama sphaerocarpa, in a semi- Vilà, M., Espinar, J. L., Hejda, M., Hulme, P. E., Jarošík, V., Maron, arid environment in south-east Spain. Oikos, 76, 455–464. https:// J. L., … Pyšek, P. (2011). Ecological impacts of invasive alien doi.org/10.2307/3546339 plants: A meta-analysis of their effects on species, communi- R Development Core Team. (2019). R: A language and environment for sta- ties and ecosystems. Ecology Letters, 14, 702–708. https://doi. tistical computing. Version 3.6.0. Vienna, Austria: R Foundation for org/10.1111/j.1461-0248.2011.01628.x Statistical Computing. Retrieved from http://www.R-project.org Waller, L. P., Allen, W. J., Barratt, B. I. P., Condron, L. M., França, F. M., Rasband, W. S. (2018). ImageJ. Bethesda, MD: U.S. National Institutes of Hunt, J. E., … Dickie, I. A. (2020). Biotic interactions drive ecosys- Health. Retrieved from https://imagej.nih.gov/ij/, 1997–2018. tem responses to plant invaders. Science, 368, 967–972. https://doi. Richardson, D. M., Pyšek, P., Rejmánek, M., Barbour, M. G., Panetta, F. org/10.1126/science.aba2225 D., & West, C. J. (2000). Naturalization and invasion of alien plants: Waller, L. P., Callaway, R. M., Klironomos, J. N., Ortega, Y. K., & Maron, Concepts and definitions. Diversity and Distributions, 6, 93–107. J. L. (2016). Reduced mycorrhizal responsiveness leads to increased https://doi.org/10.1046/j.1472-4642.2000.00083.x competitive tolerance in an invasive exotic plant. Journal of Ecology, Sauve, A. M. C., Thébault, E., Pocock, M. J. O., & Fontaine, C. (2016). 104, 1599–1607. https://doi.org/10.1111/1365-2745.12641 How plants connect pollination and herbivory networks and their Yang, S., Ferrari, M. J., & Shea, K. (2011). Pollinator behavior me- contribution to community stability. Ecology, 97, 908–917. https:// diates negative interactions between two congeneric invasive doi.org/10.1890/15-0132.1 plant species. The American Naturalist, 177, 110–118. https://doi. Simberloff, D., & Von Holle, B. (1999). Positive interactions of nonindig- org/10.1086/657433 enous species: Invasional meltdown? Biological Invasions, 1, 21–32. https://doi.org/10.1023/A:101008632 Spencer, J., van Heezik, Y., Seddon, P. J., & Barratt, B. I. P. (2017). Synergy SUPPORTING INFORMATION between two invasive species, redback spiders and rabbits, threaten the endangered Cromwell chafer beetle. Biological Invasions, 19, Additional supporting information may be found online in the 1379–1387. https://doi.org/10.1007/s10530-016-1352-6 Supporting Information section. Stotz, G. C., Cahill, J. F., Bennett, J. A., Carlyle, C. N., Bork, E. W., Askarizadeh, D., … Fraser, L. H. (2020). Not a melting pot: Plant species aggregate in their non-native range. Global Ecology and How to cite this article: Allen WJ, Wainer R, Tylianakis JM, Biogeography, 29, 482–490. https://doi.org/10.1111/geb.13046 et al. Community-level direct and indirect impacts of an Tack, A. J. M., Gripenberg, S., & Roslin, T. (2011). Can we predict indi- invasive plant favour exotic over native species. J Ecol. rect interactions from quantitative food webs? – An experimental approach. Journal of Animal Ecology, 80, 108–118. https://doi. 2020;00:1–12. https://doi.org/10.1111/1365-2745.13433 org/10.1111/j.1365-2656.2010.01744.x